Re: [dpdk-dev] [PATCH v4 02/16] net/mrvl: add mrvl net pmd driver skeleton
+1 with Thomas, see below,
Le 12/10/2017 à 08:51, Tomasz Duszynski a écrit :
What is MUSDK_DMA_MEMSIZE?
If the value cannot change, it must be a constant in the code.
If it can change, it should be a run-time driver option.
It's up to the user what MUSDK_DMA_MEMSIZE is going to be. Currently it's
set to value that should work it all cases.
Except that, MUSDK_DMA_MEMSIZE is used as synchronization point for net
and crypto (on condition they are used together i.e ipsec-secgw).
Suppose we have two different MUSDK_DMA_MEMSIZE defined for net/crypto then
dma memsize allocated will depend on driver probing sequence which might
confuse user.
It does not make sense,
+ /*
+* ret == -EEXIST is correct, it means DMA
+* has been already initialized (by another PMD).
+*/
+ ret = mv_sys_dma_mem_init(RTE_MRVL_MUSDK_DMA_MEMSIZE
int mv_sys_dma_mem_init(u64 size)
{
struct sys_dma *i_sys_dma;
int err;
if (sys_dma) {
pr_err("Dma object already exits.\n");
return -EEXIST;
}
So, I do not understand why you cannot add some checks into the drivers
to assert that users must have set the same value for both when calling:
ret = mv_sys_dma_mem_init(my_best_size);
maybe, you need to fix and improve musdk first to avoid DPDK from
getting such compilation issues.
best regards,
Vincent
Re: [dpdk-dev] [PATCH] doc: add build steps to mrvl NIC guide
12/10/2017 08:28, Tomasz Duszynski: > On Thu, Oct 12, 2017 at 04:37:45AM +0200, Thomas Monjalon wrote: > > Show how to compile MUSDK and enable compilation of the mrvl PMD. > > > > The build test tool is also updated to support this new PMD. > > > > Signed-off-by: Thomas Monjalon > > --- > > devtools/test-build.sh | 4 > > doc/guides/nics/mrvl.rst | 15 +++ > > 2 files changed, 19 insertions(+) > > > > diff --git a/devtools/test-build.sh b/devtools/test-build.sh > > index c6dfaf0a8..4d7eaa499 100755 > > --- a/devtools/test-build.sh > > +++ b/devtools/test-build.sh > > @@ -47,6 +47,7 @@ default_path=$PATH > > # - DPDK_DEP_ZLIB (y/[n]) > > # - DPDK_MAKE_JOBS (int) > > # - DPDK_NOTIFY (notify-send) > > +# - LIBMUSDK_PATH > > # - LIBSSO_SNOW3G_PATH > > # - LIBSSO_KASUMI_PATH > > # - LIBSSO_ZUC_PATH > > @@ -129,6 +130,7 @@ reset_env () > > unset DPDK_DEP_ZLIB > > unset AESNI_MULTI_BUFFER_LIB_PATH > > unset ARMV8_CRYPTO_LIB_PATH > > + unset LIBMUSDK_PATH > > unset LIBSSO_SNOW3G_PATH > > unset LIBSSO_KASUMI_PATH > > unset LIBSSO_ZUC_PATH > > @@ -169,6 +171,8 @@ config () # > > sed -ri 's,(RESOURCE_TAR=)n,\1y,' $1/.config > > test "$DPDK_DEP_MOFED" != y || \ > > sed -ri 's,(MLX._PMD=)n,\1y,' $1/.config > > + test -z "$LIBMUSDK_PATH" || \ > > + sed -ri 's,(MRVL_PMD=)n,\1y,' $1/.config > > test-build.sh modifications come with crypto-mrvl patches. The only > difference is that PMD_MRVL_CRYPTO is enabled instead of MRVL_PMD. Thus > I don't think it will apply after applying crypto patches. OK I will adapt with crypto patches. > > --- a/doc/guides/nics/mrvl.rst > > +++ b/doc/guides/nics/mrvl.rst > > @@ -221,9 +221,24 @@ Building DPDK > > Driver needs precompiled MUSDK library during compilation. Please consult > > ``doc/musdk_get_started.txt`` for the detailed build instructions. > > > > +.. code-block:: console > > + > > + export CROSS_COMPILE=/bin/aarch64-linux-gnu- > > + ./bootstrap > > + ./configure --enable-bpool-dma=64 > > + make install > > I personally build MUSDK as follows: > > export > CROSS_COMPILE=/home/tdu/workspace/gcc-linaro-5.4.1-2017.01-x86_64_aarch64-linux-gnu/bin/aarch64-linux-gnu- > > make clean > ./configure \ > --prefix=$(pwd)/musdk-install-dir \ > --enable-bpool-dma=64 \ > --enable-sam \ > --disable-shared > > make install > > '--enable-sam' is of course optional if you're not going to use crypto > engine. > > MUSDK ends up in musdk-install-dir then. Otherwise you'll need extra > permissions to install to /usr/local. Of course its up to you where > would you like to have it installed. No, the default is to install in usr/local inside musdk directory. > The reason I tend to add '--disable-shared' is that during build > DPDK will suck in static libraries and then later on I don't have to > install MUSDK library on the development board. > > > + > > Before the DPDK build process the environmental variable ``LIBMUSDK_PATH`` > > with > > the path to the MUSDK installation directory needs to be exported. > > > > +.. code-block:: console > > + > > + export LIBMUSDK_PATH=/usr/local > > + export CROSS=aarch64-linux-gnu- > > + make config T=arm64-armv8a-linuxapp-gcc > > + sed -ri 's,(MRVL_PMD=)n,\1y,' build/.config > > + make > > + [...] > > Anyway patch looks good. Thanks. > Acked-by: Tomasz Duszynski > > -- > - Tomasz Duszyński Thanks
Re: [dpdk-dev] [dpdk-stable] [PATCH v2] crypto: use snprintf return value correctly
> -Original Message-
> From: stable [mailto:stable-boun...@dpdk.org] On Behalf Of Tomasz
> Duszynski
> Sent: Wednesday, October 11, 2017 12:05 PM
> To: dev@dpdk.org
> Cc: Tomasz Duszynski ; sta...@dpdk.org
> Subject: [dpdk-stable] [PATCH v2] crypto: use snprintf return value correctly
>
> snprintf return value is the length of was encoded into destination array
> excluding '\0'. Thus return value equal to the length of the destination array
> or more means truncation.
>
> This commit fixes improper use of the return value.
>
> Fixes: eec136f3c54f ("aesni_gcm: add driver for AES-GCM crypto
> operations")
> Fixes: 924e84f87306 ("aesni_mb: add driver for multi buffer based crypto")
> Fixes: 0f548b50a160 ("crypto/aesni_mb: process crypto op on dequeue")
> Fixes: 169ca3db550c ("crypto/armv8: add PMD optimized for ARMv8
> processors")
> Fixes: 2773c86d061a ("crypto/kasumi: add driver for KASUMI library")
> Fixes: 94b0ad8e0aa5 ("null_crypto: add driver for null crypto operations")
> Fixes: d61f70b4c918 ("crypto/libcrypto: add driver for OpenSSL library")
> Fixes: 3aafc423cf4d ("snow3g: add driver for SNOW 3G library")
> Fixes: cf7685d68f00 ("crypto/zuc: add driver for ZUC library")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Tomasz Duszynski
> Acked-by: Bruce Richardson
Applied to dpdk-next-crypto.
Thanks,
Pablo
[dpdk-dev] [PATCH v4 00/16] Move PCI away from the EAL
Hi all, Here is a new version of the PCI bus move out of the EAL. The EAL PCI implementation is divided in two parts: - librte_pci: library offering helpers to handle PCI objects - librte_bus_pci: bus driver for PCI devices This allows other libraries / tools to use PCI elements (location, mappings, parsing operations, etc) without forcing a dependency on a bus driver. The latter should not have to export helpers that others might need. It is focused on defining the rte_pci_device, rte_pci_driver objects and their handling. The cryptodev library has hard dependencies on rte_pci_devices (used by generic probe function). Other similar libs (ether and eventdev) avoided the issue by inlining such functions and expecting users to include the relevant headers once the PCI bus has already been built. After review from Declan, he proposed to submit a patch removing this dependency. Once this patch is submitted, the relevant commit will be dropped from this patchset. v2: + Made rte_eal_using_phys_addrs common to both linux and bsd interfaces. + Added documentation of EAL API changes in release note. + Fixed a few rebase-related mistakes. + Fixed parallel build race condition reported by Luca Boccassi. + Grouped together commits breaking compilation: -> pci: introduce PCI lib and bus -> lib: include rte_bus_pci -> drivers: include rte_bus_pci -> test: include rte_bus_pci -> app/testpmd: include rte_bus_pci -> cryptodev: move PCI specific helpers to drivers/crypto Until all of them have been applied, compilation is broken. I am currently wondering whether merging some of them might be sensible. + Not included in this series: Several filesystem-related functions are currently private to the EAL and directly linked. This is not good, but the solution seems to be to have a new lib offering an FS abstraction. This seems an overreach for this patchset and should probably come in a second step. v3: + Fixed .map versioning + merged one commit breaking the build into the main commit moving code around. Other such commits are still present, as they only break specific subsystems (lib, drivers, apps, cryptodev). Merging them all within the one main commit does not seem right. As such, build is still broken from * pci: introduce PCI lib and bus until * cryptodev: move PCI specific helpers to drivers/crypto v4: + Rebased unto master, with new PCI functionalities integrated. + Removed the exposition of private EAL functions. While one commit did deal with this for one function, the issue is more widespread and should be fixed in a more generic way. + Introduced new PCI address parsing function, deprecating the old ones. + Fix conflict with bonding PMD regarding pci_addr_cmp function name. Compilation is still broken from * pci: introduce PCI lib and bus until * net/bonding: use local prefix for local function Gaetan Rivet (16): eal: include debug header in bus source ethdev: remove useless PCI dependency pci: introduce PCI lib and bus lib: include PCI bus header drivers: include PCI bus header test: include PCI bus header app/testpmd: include PCI bus header cryptodev: move PCI specific helpers to drivers/crypto net/bonding: use local prefix for local function pci: avoid inlining functions pci: avoid over-complicated macro pci: deprecate misnamed functions pci: introduce PCI address parsing function pci: make specialized parsing functions private bus/pci: use new PCI addr parsing function doc: add notes on EAL PCI API update app/test-pmd/testpmd.h | 1 + config/common_base | 10 + doc/guides/rel_notes/deprecation.rst | 10 + doc/guides/rel_notes/release_17_11.rst | 28 + drivers/bus/Makefile | 2 + drivers/bus/pci/Makefile | 59 ++ drivers/bus/pci/bsd/Makefile | 32 + drivers/bus/pci/bsd/pci.c| 681 +++ drivers/bus/pci/include/rte_bus_pci.h| 415 drivers/bus/pci/linux/Makefile | 37 + drivers/bus/pci/linux/pci.c | 821 +++ drivers/bus/pci/linux/pci_init.h | 97 +++ drivers/bus/pci/linux/pci_uio.c | 568 drivers/bus/pci/linux/pci_vfio.c | 757 + drivers/bus/pci/linux/vfio_mp_sync.c | 426 drivers/bus/pci/pci_common.c | 535 +++ drivers/bus/pci/pci_common_uio.c | 235 +++ drivers/bus/pci/private.h| 174 + drivers/bus/pci/rte_bus_pci_version.map | 21 + drivers/crypto/Makefile | 4 +- drivers/crypto/pci/Makefile
[dpdk-dev] [PATCH v4 01/16] eal: include debug header in bus source
This header is included through rte_pci.h, which will be removed once the PCI bus is moved out of the EAL. Signed-off-by: Gaetan Rivet --- lib/librte_eal/common/eal_common_bus.c | 1 + 1 file changed, 1 insertion(+) diff --git a/lib/librte_eal/common/eal_common_bus.c b/lib/librte_eal/common/eal_common_bus.c index c3c77f4..3c66a02 100644 --- a/lib/librte_eal/common/eal_common_bus.c +++ b/lib/librte_eal/common/eal_common_bus.c @@ -35,6 +35,7 @@ #include #include +#include #include "eal_private.h" -- 2.1.4
[dpdk-dev] [PATCH v4 02/16] ethdev: remove useless PCI dependency
Signed-off-by: Gaetan Rivet --- lib/librte_ether/rte_ethdev.c | 1 - 1 file changed, 1 deletion(-) diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c index 9460161..bb08204 100644 --- a/lib/librte_ether/rte_ethdev.c +++ b/lib/librte_ether/rte_ethdev.c @@ -47,7 +47,6 @@ #include #include #include -#include #include #include #include -- 2.1.4
[dpdk-dev] [PATCH v4 04/16] lib: include PCI bus header
Devices and drivers are now defined within the bus-specific PCI header.
Update the libraries, as structuraly unsound as it may be.
Signed-off-by: Gaetan Rivet
---
lib/librte_ether/rte_ethdev_pci.h | 1 +
lib/librte_eventdev/rte_eventdev_pmd_pci.h | 1 +
2 files changed, 2 insertions(+)
diff --git a/lib/librte_ether/rte_ethdev_pci.h
b/lib/librte_ether/rte_ethdev_pci.h
index 56b1072..722075e 100644
--- a/lib/librte_ether/rte_ethdev_pci.h
+++ b/lib/librte_ether/rte_ethdev_pci.h
@@ -36,6 +36,7 @@
#include
#include
+#include
#include
/**
diff --git a/lib/librte_eventdev/rte_eventdev_pmd_pci.h
b/lib/librte_eventdev/rte_eventdev_pmd_pci.h
index b6bd731..ade32b5 100644
--- a/lib/librte_eventdev/rte_eventdev_pmd_pci.h
+++ b/lib/librte_eventdev/rte_eventdev_pmd_pci.h
@@ -50,6 +50,7 @@ extern "C" {
#include
#include
#include
+#include
#include "rte_eventdev_pmd.h"
--
2.1.4
[dpdk-dev] [PATCH v4 05/16] drivers: include PCI bus header
Devices and drivers are now defined within the bus-specific PCI header. Update drivers. Signed-off-by: Gaetan Rivet --- drivers/bus/pci/bsd/pci.c | 1 + drivers/bus/pci/linux/pci.c| 1 + drivers/bus/pci/linux/pci_uio.c| 1 + drivers/bus/pci/linux/pci_vfio.c | 1 + drivers/bus/pci/linux/vfio_mp_sync.c | 1 + drivers/bus/pci/pci_common.c | 1 + drivers/bus/pci/pci_common_uio.c | 1 + drivers/bus/pci/private.h | 1 + drivers/crypto/qat/qat_qp.c| 1 + drivers/event/octeontx/ssovf_probe.c | 1 + drivers/mempool/octeontx/octeontx_fpavf.c | 2 +- drivers/net/ark/ark_ethdev.c | 1 + drivers/net/avp/avp_ethdev.c | 2 ++ drivers/net/bnxt/bnxt.h| 1 + drivers/net/bonding/rte_eth_bond_args.c| 1 + drivers/net/cxgbe/base/adapter.h | 1 + drivers/net/cxgbe/cxgbe_ethdev.c | 1 + drivers/net/e1000/em_ethdev.c | 1 + drivers/net/e1000/igb_ethdev.c | 1 + drivers/net/e1000/igb_pf.c | 1 + drivers/net/ena/ena_ethdev.h | 1 + drivers/net/enic/base/vnic_dev.h | 4 +++- drivers/net/enic/enic_ethdev.c | 1 + drivers/net/enic/enic_main.c | 1 + drivers/net/i40e/i40e_ethdev.c | 1 + drivers/net/i40e/i40e_ethdev_vf.c | 1 + drivers/net/ixgbe/ixgbe_ethdev.c | 1 + drivers/net/ixgbe/ixgbe_ethdev.h | 1 + drivers/net/mlx4/mlx4_ethdev.c | 2 +- drivers/net/mlx5/mlx5.c| 1 + drivers/net/mlx5/mlx5_ethdev.c | 1 + drivers/net/nfp/nfp_nfpu.c | 2 +- drivers/net/nfp/nfp_nfpu.h | 2 +- drivers/net/octeontx/base/octeontx_pkivf.c | 2 +- drivers/net/octeontx/base/octeontx_pkovf.c | 2 +- drivers/net/sfc/sfc.h | 1 + drivers/net/sfc/sfc_ethdev.c | 1 + drivers/net/thunderx/nicvf_ethdev.c| 1 + drivers/net/virtio/virtio_ethdev.c | 1 + drivers/net/virtio/virtio_pci.h| 1 + drivers/net/vmxnet3/vmxnet3_ethdev.c | 1 + 41 files changed, 44 insertions(+), 7 deletions(-) diff --git a/drivers/bus/pci/bsd/pci.c b/drivers/bus/pci/bsd/pci.c index 599519a..753d914 100644 --- a/drivers/bus/pci/bsd/pci.c +++ b/drivers/bus/pci/bsd/pci.c @@ -57,6 +57,7 @@ #include #include #include +#include #include #include #include diff --git a/drivers/bus/pci/linux/pci.c b/drivers/bus/pci/linux/pci.c index c1fcc56..422579f 100644 --- a/drivers/bus/pci/linux/pci.c +++ b/drivers/bus/pci/linux/pci.c @@ -37,6 +37,7 @@ #include #include #include +#include #include #include #include diff --git a/drivers/bus/pci/linux/pci_uio.c b/drivers/bus/pci/linux/pci_uio.c index d3f0556..940fa21 100644 --- a/drivers/bus/pci/linux/pci_uio.c +++ b/drivers/bus/pci/linux/pci_uio.c @@ -47,6 +47,7 @@ #include #include +#include #include #include #include diff --git a/drivers/bus/pci/linux/pci_vfio.c b/drivers/bus/pci/linux/pci_vfio.c index e69e4d3..855e2b9 100644 --- a/drivers/bus/pci/linux/pci_vfio.c +++ b/drivers/bus/pci/linux/pci_vfio.c @@ -42,6 +42,7 @@ #include #include +#include #include #include diff --git a/drivers/bus/pci/linux/vfio_mp_sync.c b/drivers/bus/pci/linux/vfio_mp_sync.c index dd24e86..d07687f 100644 --- a/drivers/bus/pci/linux/vfio_mp_sync.c +++ b/drivers/bus/pci/linux/vfio_mp_sync.c @@ -50,6 +50,7 @@ #include #include +#include #include #include diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c index 7b68e99..cd95d1a 100644 --- a/drivers/bus/pci/pci_common.c +++ b/drivers/bus/pci/pci_common.c @@ -45,6 +45,7 @@ #include #include #include +#include #include #include #include diff --git a/drivers/bus/pci/pci_common_uio.c b/drivers/bus/pci/pci_common_uio.c index 4365660..544c606 100644 --- a/drivers/bus/pci/pci_common_uio.c +++ b/drivers/bus/pci/pci_common_uio.c @@ -40,6 +40,7 @@ #include #include +#include #include #include #include diff --git a/drivers/bus/pci/private.h b/drivers/bus/pci/private.h index 7ff1fc4..fdc2c81 100644 --- a/drivers/bus/pci/private.h +++ b/drivers/bus/pci/private.h @@ -37,6 +37,7 @@ #include #include #include +#include struct rte_pci_driver; struct rte_pci_device; diff --git a/drivers/crypto/qat/qat_qp.c b/drivers/crypto/qat/qat_qp.c index 34f75ca..8489662 100644 --- a/drivers/crypto/qat/qat_qp.c +++ b/drivers/crypto/qat/qat_qp.c @@ -37,6 +37,7 @@ #include #include #include +#include #include #include diff --git a/drivers/event/octeontx/ssovf_probe.c b/drivers/event/octeontx/ssovf_probe.c index e1c0c6d..1cac4bc 100644 --- a/drivers/event/octeontx/ssovf_probe.c +++ b/drivers/event/octeontx/ssovf_probe.c @@ -35,6 +35,7 @@ #include #include #include +#include #include "ssovf_evdev.h" diff --git a/drivers/mempool/octeontx/octeontx_fpavf.c
[dpdk-dev] [PATCH v4 06/16] test: include PCI bus header
Devices and drivers are now defined within the bus-specific PCI header. Update test applications. Signed-off-by: Gaetan Rivet --- test/test/test_kni.c| 1 + test/test/virtual_pmd.c | 1 + 2 files changed, 2 insertions(+) diff --git a/test/test/test_kni.c b/test/test/test_kni.c index 2450c9f..b956727 100644 --- a/test/test/test_kni.c +++ b/test/test/test_kni.c @@ -42,6 +42,7 @@ #include #include #include +#include #include #include diff --git a/test/test/virtual_pmd.c b/test/test/virtual_pmd.c index 09daf6c..b57a949 100644 --- a/test/test/virtual_pmd.c +++ b/test/test/virtual_pmd.c @@ -34,6 +34,7 @@ #include #include #include +#include #include #include #include -- 2.1.4
[dpdk-dev] [PATCH v4 10/16] pci: avoid inlining functions
Parsing operations should not happen in performance critical sections.
Headers should not propose implementations unless duly required.
Signed-off-by: Gaetan Rivet
---
lib/librte_pci/include/rte_pci.h | 69 --
lib/librte_pci/rte_pci.c | 65 +++
lib/librte_pci/rte_pci_version.map | 4 +++
3 files changed, 75 insertions(+), 63 deletions(-)
diff --git a/lib/librte_pci/include/rte_pci.h b/lib/librte_pci/include/rte_pci.h
index 2647568..fe4d411 100644
--- a/lib/librte_pci/include/rte_pci.h
+++ b/lib/librte_pci/include/rte_pci.h
@@ -134,19 +134,6 @@ struct mapped_pci_resource {
/** mapped pci device list */
TAILQ_HEAD(mapped_pci_res_list, mapped_pci_resource);
-/**< Internal use only - Macro used by pci addr parsing functions **/
-#define GET_PCIADDR_FIELD(in, fd, lim, dlm) \
-do { \
- unsigned long val; \
- char *end; \
- errno = 0; \
- val = strtoul((in), &end, 16); \
- if (errno != 0 || end[0] != (dlm) || val > (lim)) \
- return -EINVAL; \
- (fd) = (typeof (fd))val;\
- (in) = end + 1; \
-} while(0)
-
/**
* Utility function to produce a PCI Bus-Device-Function value
* given a string representation. Assumes that the BDF is provided without
@@ -160,15 +147,7 @@ do {
\
* @return
* 0 on success, negative on error.
*/
-static inline int
-eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr)
-{
- dev_addr->domain = 0;
- GET_PCIADDR_FIELD(input, dev_addr->bus, UINT8_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->devid, UINT8_MAX, '.');
- GET_PCIADDR_FIELD(input, dev_addr->function, UINT8_MAX, 0);
- return 0;
-}
+int eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr);
/**
* Utility function to produce a PCI Bus-Device-Function value
@@ -182,16 +161,7 @@ eal_parse_pci_BDF(const char *input, struct rte_pci_addr
*dev_addr)
* @return
* 0 on success, negative on error.
*/
-static inline int
-eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr)
-{
- GET_PCIADDR_FIELD(input, dev_addr->domain, UINT16_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->bus, UINT8_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->devid, UINT8_MAX, '.');
- GET_PCIADDR_FIELD(input, dev_addr->function, UINT8_MAX, 0);
- return 0;
-}
-#undef GET_PCIADDR_FIELD
+int eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr);
/**
* Utility function to write a pci device name, this device name can later be
@@ -205,17 +175,9 @@ eal_parse_pci_DomBDF(const char *input, struct
rte_pci_addr *dev_addr)
* @param size
* The output buffer size
*/
-static inline void
-rte_pci_device_name(const struct rte_pci_addr *addr,
- char *output, size_t size)
-{
- RTE_VERIFY(size >= PCI_PRI_STR_SIZE);
- RTE_VERIFY(snprintf(output, size, PCI_PRI_FMT,
- addr->domain, addr->bus,
- addr->devid, addr->function) >= 0);
-}
+void rte_pci_device_name(const struct rte_pci_addr *addr,
+char *output, size_t size);
-/* Compare two PCI device addresses. */
/**
* Utility function to compare two PCI device addresses.
*
@@ -228,27 +190,8 @@ rte_pci_device_name(const struct rte_pci_addr *addr,
* Positive on addr is greater than addr2.
* Negative on addr is less than addr2, or error.
*/
-static inline int
-rte_eal_compare_pci_addr(const struct rte_pci_addr *addr,
-const struct rte_pci_addr *addr2)
-{
- uint64_t dev_addr, dev_addr2;
-
- if ((addr == NULL) || (addr2 == NULL))
- return -1;
-
- dev_addr = ((uint64_t)addr->domain << 24) |
- (addr->bus << 16) | (addr->devid << 8) | addr->function;
- dev_addr2 = ((uint64_t)addr2->domain << 24) |
- (addr2->bus << 16) | (addr2->devid << 8) | addr2->function;
-
- if (dev_addr > dev_addr2)
- return 1;
- else if (dev_addr < dev_addr2)
- return -1;
- else
- return 0;
-}
+int rte_eal_compare_pci_addr(const struct rte_pci_addr *addr,
+const struct rte_pci_addr *addr2);
/**
* Map a particular resource from a file.
diff --git a/lib/librte_pci/rte_pci.c b/lib/librte_pci/rte_pci.c
index 9dfdd3f..8584b55 100644
--- a/lib/librte_pci/rte_pci.c
+++ b/lib/librte_pci/rte_pci.c
@@ -53,6 +53,71 @@
#include "rte_pci.h"
+/* Macro used by pci addr parsing functions. **/
+#d
[dpdk-dev] [PATCH v4 09/16] net/bonding: use local prefix for local function
Signed-off-by: Gaetan Rivet
---
drivers/net/bonding/rte_eth_bond_args.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/bonding/rte_eth_bond_args.c
b/drivers/net/bonding/rte_eth_bond_args.c
index dea0265..e816da3 100644
--- a/drivers/net/bonding/rte_eth_bond_args.c
+++ b/drivers/net/bonding/rte_eth_bond_args.c
@@ -90,7 +90,7 @@ find_port_id_by_dev_name(const char *name)
}
static inline int
-pci_addr_cmp(const struct rte_device *dev, const void *_pci_addr)
+bond_pci_addr_cmp(const struct rte_device *dev, const void *_pci_addr)
{
struct rte_pci_device *pdev;
const struct rte_pci_addr *paddr = _pci_addr;
@@ -119,7 +119,7 @@ parse_port_id(const char *port_str)
/* try parsing as pci address, physical devices */
if (pci_bus->parse(port_str, &dev_addr) == 0) {
- dev = pci_bus->find_device(NULL, pci_addr_cmp, &dev_addr);
+ dev = pci_bus->find_device(NULL, bond_pci_addr_cmp, &dev_addr);
if (dev == NULL) {
RTE_LOG(ERR, PMD, "unable to find PCI device\n");
return -1;
--
2.1.4
[dpdk-dev] [PATCH v4 07/16] app/testpmd: include PCI bus header
Devices and drivers are now defined within the bus-specific PCI header. Update applications. Signed-off-by: Gaetan Rivet --- app/test-pmd/testpmd.h | 1 + 1 file changed, 1 insertion(+) diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h index 0b8e624..b05c8a0 100644 --- a/app/test-pmd/testpmd.h +++ b/app/test-pmd/testpmd.h @@ -35,6 +35,7 @@ #define _TESTPMD_H_ #include +#include #include #include -- 2.1.4
[dpdk-dev] [PATCH v4 08/16] cryptodev: move PCI specific helpers to drivers/crypto
Those helpers rely on the PCI bus driver implementation. Other similar libraries relied on the bus-specifics being handled in inlined functions, to be compiled on demand by drivers, once the proper PCI dependency has been settled. This seems unsafe. Move the PCI-specific helpers out of the lib directory to the drivers/crypto directory, properly following the dependency hierarchy. Signed-off-by: Gaetan Rivet --- drivers/crypto/Makefile | 4 +- drivers/crypto/pci/Makefile | 52 + drivers/crypto/pci/rte_cryptodev_pci.c | 128 +++ drivers/crypto/pci/rte_cryptodev_pci.h | 94 + drivers/crypto/pci/rte_cryptodev_pci_version.map | 7 ++ lib/librte_cryptodev/Makefile| 1 - lib/librte_cryptodev/rte_cryptodev_pci.h | 92 lib/librte_cryptodev/rte_cryptodev_pmd.c | 94 - lib/librte_cryptodev/rte_cryptodev_version.map | 2 - 9 files changed, 284 insertions(+), 190 deletions(-) create mode 100644 drivers/crypto/pci/Makefile create mode 100644 drivers/crypto/pci/rte_cryptodev_pci.c create mode 100644 drivers/crypto/pci/rte_cryptodev_pci.h create mode 100644 drivers/crypto/pci/rte_cryptodev_pci_version.map delete mode 100644 lib/librte_cryptodev/rte_cryptodev_pci.h diff --git a/drivers/crypto/Makefile b/drivers/crypto/Makefile index 7a719b9..cfd6cb6 100644 --- a/drivers/crypto/Makefile +++ b/drivers/crypto/Makefile @@ -33,6 +33,8 @@ include $(RTE_SDK)/mk/rte.vars.mk core-libs := librte_eal librte_mbuf librte_mempool librte_ring librte_cryptodev +DIRS-$(CONFIG_RTE_LIBRTE_PCI_BUS) += pci +DEPDIRS-pci = $(core-libs) DIRS-$(CONFIG_RTE_LIBRTE_PMD_AESNI_GCM) += aesni_gcm DEPDIRS-aesni_gcm = $(core-libs) DIRS-$(CONFIG_RTE_LIBRTE_PMD_AESNI_MB) += aesni_mb @@ -42,7 +44,7 @@ DEPDIRS-armv8 = $(core-libs) DIRS-$(CONFIG_RTE_LIBRTE_PMD_OPENSSL) += openssl DEPDIRS-openssl = $(core-libs) DIRS-$(CONFIG_RTE_LIBRTE_PMD_QAT) += qat -DEPDIRS-qat = $(core-libs) +DEPDIRS-qat = $(core-libs) librte_cryptodev_pci DIRS-$(CONFIG_RTE_LIBRTE_PMD_CRYPTO_SCHEDULER) += scheduler DEPDIRS-scheduler = $(core-libs) librte_kvargs librte_reorder DIRS-$(CONFIG_RTE_LIBRTE_PMD_SNOW3G) += snow3g diff --git a/drivers/crypto/pci/Makefile b/drivers/crypto/pci/Makefile new file mode 100644 index 000..da819f2 --- /dev/null +++ b/drivers/crypto/pci/Makefile @@ -0,0 +1,52 @@ +# BSD LICENSE +# +# Copyright(c) 2017 6WIND S.A. All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of 6WIND S.A. nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +include $(RTE_SDK)/mk/rte.vars.mk + +# library name +LIB = librte_cryptodev_pci.a + +# library version +LIBABIVER := 1 + +# build flags +CFLAGS += -O3 +CFLAGS += $(WERROR_FLAGS) + +# library source files +SRCS-y += rte_cryptodev_pci.c + +# export include files +SYMLINK-y-include += rte_cryptodev_pci.h + +# versioning export map +EXPORT_MAP := rte_cryptodev_pci_version.map + +include $(RTE_SDK)/mk/rte.lib.mk diff --git a/drivers/crypto/pci/rte_cryptodev_pci.c b/drivers/crypto/pci/rte_cryptodev_pci.c new file mode 100644 index 000..a2a1366 --- /dev/null +++ b/drivers/crypto/pci/rte_cryptodev_pci.c @@ -0,0 +1,128 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2017 6WIND S.A. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * *
[dpdk-dev] [PATCH v4 12/16] pci: deprecate misnamed functions
Rename misnamed functions and describe the change in a deprecation
notice.
Signed-off-by: Gaetan Rivet
---
doc/guides/rel_notes/deprecation.rst | 10 ++
lib/librte_pci/include/rte_pci.h | 63
lib/librte_pci/rte_pci.c | 26 +++
lib/librte_pci/rte_pci_version.map | 4 +++
4 files changed, 103 insertions(+)
diff --git a/doc/guides/rel_notes/deprecation.rst
b/doc/guides/rel_notes/deprecation.rst
index 4e4d97b..ef2264f 100644
--- a/doc/guides/rel_notes/deprecation.rst
+++ b/doc/guides/rel_notes/deprecation.rst
@@ -98,3 +98,13 @@ Deprecation Notices
The non-"do-sig" versions of the hash tables will be removed
(including the ``signature_offset`` parameter)
and the "do-sig" versions renamed accordingly.
+
+* pci: Several exposed functions are misnamed.
+ The following functions are deprecated starting from v17.11 and are replaced:
+
+ - ``eal_parse_pci_BDF`` replaced by ``pci_parse_BDF``
+ - ``eal_parse_pci_DomBDF`` replaced by ``pci_parse_DomBDF``
+ - ``rte_eal_compare_pci_addr`` replaced by ``pci_addr_cmp``
+ - ``rte_pci_device_name`` replaced by ``pci_device_name``
+
+ The functions are only renamed. Their behavior is not affected.
diff --git a/lib/librte_pci/include/rte_pci.h b/lib/librte_pci/include/rte_pci.h
index fe4d411..38cbdb3 100644
--- a/lib/librte_pci/include/rte_pci.h
+++ b/lib/librte_pci/include/rte_pci.h
@@ -135,6 +135,7 @@ struct mapped_pci_resource {
TAILQ_HEAD(mapped_pci_res_list, mapped_pci_resource);
/**
+ * @deprecated
* Utility function to produce a PCI Bus-Device-Function value
* given a string representation. Assumes that the BDF is provided without
* a domain prefix (i.e. domain returned is always 0)
@@ -151,6 +152,22 @@ int eal_parse_pci_BDF(const char *input, struct
rte_pci_addr *dev_addr);
/**
* Utility function to produce a PCI Bus-Device-Function value
+ * given a string representation. Assumes that the BDF is provided without
+ * a domain prefix (i.e. domain returned is always 0)
+ *
+ * @param input
+ * The input string to be parsed. Should have the format XX:XX.X
+ * @param dev_addr
+ * The PCI Bus-Device-Function address to be returned.
+ * Domain will always be returned as 0
+ * @return
+ * 0 on success, negative on error.
+ */
+int pci_bdf_parse(const char *input, struct rte_pci_addr *dev_addr);
+
+/**
+ * @deprecated
+ * Utility function to produce a PCI Bus-Device-Function value
* given a string representation. Assumes that the BDF is provided including
* a domain prefix.
*
@@ -164,6 +181,21 @@ int eal_parse_pci_BDF(const char *input, struct
rte_pci_addr *dev_addr);
int eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr);
/**
+ * Utility function to produce a PCI Bus-Device-Function value
+ * given a string representation. Assumes that the BDF is provided including
+ * a domain prefix.
+ *
+ * @param input
+ * The input string to be parsed. Should have the format :XX:XX.X
+ * @param dev_addr
+ * The PCI Bus-Device-Function address to be returned
+ * @return
+ * 0 on success, negative on error.
+ */
+int pci_dbdf_parse(const char *input, struct rte_pci_addr *dev_addr);
+
+/**
+ * @deprecated
* Utility function to write a pci device name, this device name can later be
* used to retrieve the corresponding rte_pci_addr using eal_parse_pci_*
* BDF helpers.
@@ -179,6 +211,22 @@ void rte_pci_device_name(const struct rte_pci_addr *addr,
char *output, size_t size);
/**
+ * Utility function to write a pci device name, this device name can later be
+ * used to retrieve the corresponding rte_pci_addr using eal_parse_pci_*
+ * BDF helpers.
+ *
+ * @param addr
+ * The PCI Bus-Device-Function address
+ * @param output
+ * The output buffer string
+ * @param size
+ * The output buffer size
+ */
+void pci_device_name(const struct rte_pci_addr *addr,
+char *output, size_t size);
+
+/**
+ * @deprecated
* Utility function to compare two PCI device addresses.
*
* @param addr
@@ -194,6 +242,21 @@ int rte_eal_compare_pci_addr(const struct rte_pci_addr
*addr,
const struct rte_pci_addr *addr2);
/**
+ * Utility function to compare two PCI device addresses.
+ *
+ * @param addr
+ * The PCI Bus-Device-Function address to compare
+ * @param addr2
+ * The PCI Bus-Device-Function address to compare
+ * @return
+ * 0 on equal PCI address.
+ * Positive on addr is greater than addr2.
+ * Negative on addr is less than addr2, or error.
+ */
+int pci_addr_cmp(const struct rte_pci_addr *addr,
+const struct rte_pci_addr *addr2);
+
+/**
* Map a particular resource from a file.
*
* @param requested_addr
diff --git a/lib/librte_pci/rte_pci.c b/lib/librte_pci/rte_pci.c
index cbb5359..fc2fc7f 100644
--- a/lib/librte_pci/rte_pci.c
+++ b/lib/librte_pci/rte_pci.c
@@ -73,6 +73,12 @@ get_u8_pciaddr_field(const char *in,
[dpdk-dev] [PATCH v4 13/16] pci: introduce PCI address parsing function
A new single function that is able to parse all currently supported
format:
* Domain-Bus-Device-Function
*Bus-Device-Function
Signed-off-by: Gaetan Rivet
---
lib/librte_pci/include/rte_pci.h | 15 +++
lib/librte_pci/rte_pci.c | 10 ++
lib/librte_pci/rte_pci_version.map | 1 +
3 files changed, 26 insertions(+)
diff --git a/lib/librte_pci/include/rte_pci.h b/lib/librte_pci/include/rte_pci.h
index 38cbdb3..9e79557 100644
--- a/lib/librte_pci/include/rte_pci.h
+++ b/lib/librte_pci/include/rte_pci.h
@@ -256,6 +256,21 @@ int rte_eal_compare_pci_addr(const struct rte_pci_addr
*addr,
int pci_addr_cmp(const struct rte_pci_addr *addr,
const struct rte_pci_addr *addr2);
+
+/**
+ * Utility function to parse a string into a PCI location.
+ *
+ * @param str
+ * The string to parse
+ * @param addr
+ * The reference to the structure where the location
+ * is stored.
+ * @return
+ * 0 on success
+ * <0 otherwise
+ */
+int pci_addr_parse(const char *str, struct rte_pci_addr *addr);
+
/**
* Map a particular resource from a file.
*
diff --git a/lib/librte_pci/rte_pci.c b/lib/librte_pci/rte_pci.c
index fc2fc7f..7ba472c 100644
--- a/lib/librte_pci/rte_pci.c
+++ b/lib/librte_pci/rte_pci.c
@@ -171,6 +171,16 @@ pci_addr_cmp(const struct rte_pci_addr *addr,
return 0;
}
+int
+pci_addr_parse(const char *str, struct rte_pci_addr *addr)
+{
+ if (pci_bdf_parse(str, addr) == 0 ||
+ pci_dbdf_parse(str, addr) == 0)
+ return 0;
+ return -1;
+}
+
+
/* map a particular resource from a file */
void *
pci_map_resource(void *requested_addr, int fd, off_t offset, size_t size,
diff --git a/lib/librte_pci/rte_pci_version.map
b/lib/librte_pci/rte_pci_version.map
index 8d180e3..5f50f98 100644
--- a/lib/librte_pci/rte_pci_version.map
+++ b/lib/librte_pci/rte_pci_version.map
@@ -4,6 +4,7 @@ DPDK_17.11 {
eal_parse_pci_BDF;
eal_parse_pci_DomBDF;
pci_addr_cmp;
+ pci_addr_parse;
pci_bdf_parse;
pci_dbdf_parse;
pci_device_name;
--
2.1.4
[dpdk-dev] [PATCH v4 14/16] pci: make specialized parsing functions private
Do not expose the minute implementations of PCI parsing.
This leaves only the all-purpose pci_addr_parse, which is simpler to
use.
Signed-off-by: Gaetan Rivet
---
lib/librte_pci/include/rte_pci.h | 29 -
lib/librte_pci/rte_pci.c | 28 ++--
lib/librte_pci/rte_pci_version.map | 2 --
3 files changed, 14 insertions(+), 45 deletions(-)
diff --git a/lib/librte_pci/include/rte_pci.h b/lib/librte_pci/include/rte_pci.h
index 9e79557..ea0897c 100644
--- a/lib/librte_pci/include/rte_pci.h
+++ b/lib/librte_pci/include/rte_pci.h
@@ -151,21 +151,6 @@ TAILQ_HEAD(mapped_pci_res_list, mapped_pci_resource);
int eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr);
/**
- * Utility function to produce a PCI Bus-Device-Function value
- * given a string representation. Assumes that the BDF is provided without
- * a domain prefix (i.e. domain returned is always 0)
- *
- * @param input
- * The input string to be parsed. Should have the format XX:XX.X
- * @param dev_addr
- * The PCI Bus-Device-Function address to be returned.
- * Domain will always be returned as 0
- * @return
- * 0 on success, negative on error.
- */
-int pci_bdf_parse(const char *input, struct rte_pci_addr *dev_addr);
-
-/**
* @deprecated
* Utility function to produce a PCI Bus-Device-Function value
* given a string representation. Assumes that the BDF is provided including
@@ -181,20 +166,6 @@ int pci_bdf_parse(const char *input, struct rte_pci_addr
*dev_addr);
int eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr);
/**
- * Utility function to produce a PCI Bus-Device-Function value
- * given a string representation. Assumes that the BDF is provided including
- * a domain prefix.
- *
- * @param input
- * The input string to be parsed. Should have the format :XX:XX.X
- * @param dev_addr
- * The PCI Bus-Device-Function address to be returned
- * @return
- * 0 on success, negative on error.
- */
-int pci_dbdf_parse(const char *input, struct rte_pci_addr *dev_addr);
-
-/**
* @deprecated
* Utility function to write a pci device name, this device name can later be
* used to retrieve the corresponding rte_pci_addr using eal_parse_pci_*
diff --git a/lib/librte_pci/rte_pci.c b/lib/librte_pci/rte_pci.c
index 7ba472c..1307a18 100644
--- a/lib/librte_pci/rte_pci.c
+++ b/lib/librte_pci/rte_pci.c
@@ -70,13 +70,7 @@ get_u8_pciaddr_field(const char *in, void *_u8, char dlm)
return end + 1;
}
-int
-eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr)
-{
- return pci_bdf_parse(input, dev_addr);
-}
-
-int
+static int
pci_bdf_parse(const char *input, struct rte_pci_addr *dev_addr)
{
const char *in = input;
@@ -94,13 +88,7 @@ pci_bdf_parse(const char *input, struct rte_pci_addr
*dev_addr)
return 0;
}
-int
-eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr)
-{
- return pci_dbdf_parse(input, dev_addr);
-}
-
-int
+static int
pci_dbdf_parse(const char *input, struct rte_pci_addr *dev_addr)
{
const char *in = input;
@@ -125,6 +113,18 @@ pci_dbdf_parse(const char *input, struct rte_pci_addr
*dev_addr)
return 0;
}
+int
+eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr)
+{
+ return pci_bdf_parse(input, dev_addr);
+}
+
+int
+eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr)
+{
+ return pci_dbdf_parse(input, dev_addr);
+}
+
void
rte_pci_device_name(const struct rte_pci_addr *addr,
char *output, size_t size)
diff --git a/lib/librte_pci/rte_pci_version.map
b/lib/librte_pci/rte_pci_version.map
index 5f50f98..b5c9ec2 100644
--- a/lib/librte_pci/rte_pci_version.map
+++ b/lib/librte_pci/rte_pci_version.map
@@ -5,8 +5,6 @@ DPDK_17.11 {
eal_parse_pci_DomBDF;
pci_addr_cmp;
pci_addr_parse;
- pci_bdf_parse;
- pci_dbdf_parse;
pci_device_name;
pci_map_resource;
pci_unmap_resource;
--
2.1.4
[dpdk-dev] [PATCH v4 15/16] bus/pci: use new PCI addr parsing function
Signed-off-by: Gaetan Rivet --- drivers/bus/pci/pci_common.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c index cd95d1a..d7a1c05 100644 --- a/drivers/bus/pci/pci_common.c +++ b/drivers/bus/pci/pci_common.c @@ -437,8 +437,7 @@ pci_parse(const char *name, void *addr) struct rte_pci_addr pci_addr; bool parse; - parse = (eal_parse_pci_BDF(name, &pci_addr) == 0 || -eal_parse_pci_DomBDF(name, &pci_addr) == 0); + parse = (pci_addr_parse(name, &pci_addr) == 0); if (parse && addr != NULL) *out = pci_addr; return parse == false; -- 2.1.4
[dpdk-dev] [PATCH v4 11/16] pci: avoid over-complicated macro
Using a macro helps writing the code to the detriment of the reader in
this case. This is backward. Write once, read many.
The few LOCs gained is not worth the opacity of the implementation.
Signed-off-by: Gaetan Rivet
---
lib/librte_pci/rte_pci.c | 65 ++--
1 file changed, 46 insertions(+), 19 deletions(-)
diff --git a/lib/librte_pci/rte_pci.c b/lib/librte_pci/rte_pci.c
index 8584b55..cbb5359 100644
--- a/lib/librte_pci/rte_pci.c
+++ b/lib/librte_pci/rte_pci.c
@@ -53,36 +53,63 @@
#include "rte_pci.h"
-/* Macro used by pci addr parsing functions. **/
-#define GET_PCIADDR_FIELD(in, fd, lim, dlm) \
-do {\
- unsigned long val; \
- char *end; \
- errno = 0; \
- val = strtoul((in), &end, 16); \
- if (errno != 0 || end[0] != (dlm) || val > (lim)) \
- return -EINVAL; \
- (fd) = (typeof (fd))val;\
- (in) = end + 1; \
-} while(0)
+static inline const char *
+get_u8_pciaddr_field(const char *in, void *_u8, char dlm)
+{
+ unsigned long val;
+ uint8_t *u8 = _u8;
+ char *end;
+
+ errno = 0;
+ val = strtoul(in, &end, 16);
+ if (errno != 0 || end[0] != dlm || val > UINT8_MAX) {
+ errno = errno ? errno : EINVAL;
+ return NULL;
+ }
+ *u8 = (uint8_t)val;
+ return end + 1;
+}
int
eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr)
{
+ const char *in = input;
+
dev_addr->domain = 0;
- GET_PCIADDR_FIELD(input, dev_addr->bus, UINT8_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->devid, UINT8_MAX, '.');
- GET_PCIADDR_FIELD(input, dev_addr->function, UINT8_MAX, 0);
+ in = get_u8_pciaddr_field(in, &dev_addr->bus, ':');
+ if (in == NULL)
+ return -EINVAL;
+ in = get_u8_pciaddr_field(in, &dev_addr->devid, '.');
+ if (in == NULL)
+ return -EINVAL;
+ in = get_u8_pciaddr_field(in, &dev_addr->function, '\0');
+ if (in == NULL)
+ return -EINVAL;
return 0;
}
int
eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr)
{
- GET_PCIADDR_FIELD(input, dev_addr->domain, UINT16_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->bus, UINT8_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->devid, UINT8_MAX, '.');
- GET_PCIADDR_FIELD(input, dev_addr->function, UINT8_MAX, 0);
+ const char *in = input;
+ unsigned long val;
+ char *end;
+
+ errno = 0;
+ val = strtoul(in, &end, 16);
+ if (errno != 0 || end[0] != ':' || val > UINT16_MAX)
+ return -EINVAL;
+ dev_addr->domain = (uint16_t)val;
+ in = end + 1;
+ in = get_u8_pciaddr_field(in, &dev_addr->bus, ':');
+ if (in == NULL)
+ return -EINVAL;
+ in = get_u8_pciaddr_field(in, &dev_addr->devid, '.');
+ if (in == NULL)
+ return -EINVAL;
+ in = get_u8_pciaddr_field(in, &dev_addr->function, '\0');
+ if (in == NULL)
+ return -EINVAL;
return 0;
}
--
2.1.4
[dpdk-dev] [PATCH v1 0/8] Bus control framework
Probing policy was introduced in the previous release as a configuration item. It was thus added to the generic bus structure, breaking its ABI. In this release, the IOVA mode can be read from a bus to configure the EAL. This new configuration element also broke the bus ABI when it was added. As new operators had to be implemented for the probe policy item, these patches were developed to help mitigate this issue. This control framework allows to expand the rte_bus API without breaking its ABI. It is meant to be used with configuration elements that may only be valid for a few buses, while the others would remain untouched and unaware of the evolution. A central control operator is used, similarly to the working of rte_flow API in the ether layer. Each driver thus chooses to expose a set of operators relevant to its implementation. The caller is then free to use those if they are available. Both Probe mode and IOVA mode operators are implemented for the PCI bus. This patchset depends on: Move PCI away from the EAL http://dpdk.org/ml/archives/dev/2017-August/073512.html Gaetan Rivet (8): bus: rename scan policy as probe policy bus: introduce opaque control framework bus: remove probe mode configuration structure bus: add probe mode setter bus/pci: implement ctrl operator bus: add IOVA mode as a ctrl operation bus/pci: implement IOVA mode getter bus: remove redundant IOVA mode getter drivers/bus/pci/bsd/pci.c | 9 +- drivers/bus/pci/include/rte_bus_pci.h | 12 +-- drivers/bus/pci/linux/pci.c | 20 ++-- drivers/bus/pci/pci_common.c| 53 +- drivers/bus/pci/private.h | 13 +++ lib/librte_eal/bsdapp/eal/rte_eal_version.map | 1 - lib/librte_eal/common/eal_common_bus.c | 57 +-- lib/librte_eal/common/eal_common_devargs.c | 8 -- lib/librte_eal/common/eal_common_options.c | 17 +--- lib/librte_eal/common/include/rte_bus.h | 127 lib/librte_eal/linuxapp/eal/rte_eal_version.map | 1 - 11 files changed, 222 insertions(+), 96 deletions(-) -- 2.1.4
[dpdk-dev] [PATCH v4 16/16] doc: add notes on EAL PCI API update
Add a section related to EAL API changes to 17.11 release notes. Signed-off-by: Gaetan Rivet Acked-by: John McNamara --- doc/guides/rel_notes/release_17_11.rst | 28 1 file changed, 28 insertions(+) diff --git a/doc/guides/rel_notes/release_17_11.rst b/doc/guides/rel_notes/release_17_11.rst index 63b9869..ed3d6d8 100644 --- a/doc/guides/rel_notes/release_17_11.rst +++ b/doc/guides/rel_notes/release_17_11.rst @@ -232,6 +232,34 @@ API Changes By this way PMDs can return an error value in case of failure at stats getting process time. +* **PCI bus API moved outside of the EAL** + + The PCI bus previously implemented within the EAL has been moved. + A first part has been added as an RTE library providing PCI helpers to + parse device locations or other such utilities. + A second part consisting in the actual bus driver has been moved to its + proper subdirectory, without changing its functionalities. + + As such, several PCI-related functions are not proposed by the EAL anymore: + + * rte_pci_detach + * rte_pci_dump + * rte_pci_ioport_map + * rte_pci_ioport_read + * rte_pci_ioport_unmap + * rte_pci_ioport_write + * rte_pci_map_device + * rte_pci_probe + * rte_pci_probe_one + * rte_pci_read_config + * rte_pci_register + * rte_pci_scan + * rte_pci_unmap_device + * rte_pci_unregister + * rte_pci_write_config + + These functions are made available either as part of ``librte_pci`` or + ``librte_bus_pci``. ABI Changes --- -- 2.1.4
[dpdk-dev] [PATCH v1 2/8] bus: introduce opaque control framework
New configuration elements are added to the buses. They make the ABI
unstable and will continue to do so.
This new control scheme allows to add new bus operators without
breaking the ABI and by only expanding the API.
This helps having more stability in core EAL subsystems, while allowing
flexibility for future evolutions.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/common/eal_common_bus.c | 9 +++
lib/librte_eal/common/include/rte_bus.h | 46 +
2 files changed, 55 insertions(+)
diff --git a/lib/librte_eal/common/eal_common_bus.c
b/lib/librte_eal/common/eal_common_bus.c
index 3c66a02..65d7229 100644
--- a/lib/librte_eal/common/eal_common_bus.c
+++ b/lib/librte_eal/common/eal_common_bus.c
@@ -42,6 +42,13 @@
struct rte_bus_list rte_bus_list =
TAILQ_HEAD_INITIALIZER(rte_bus_list);
+static rte_bus_ctrl_t
+rte_bus_default_ctrl(enum rte_bus_ctrl_op op __rte_unused,
+enum rte_bus_ctrl_item item __rte_unused)
+{
+ return NULL;
+}
+
void
rte_bus_register(struct rte_bus *bus)
{
@@ -53,6 +60,8 @@ rte_bus_register(struct rte_bus *bus)
RTE_VERIFY(bus->find_device);
/* Buses supporting driver plug also require unplug. */
RTE_VERIFY(!bus->plug || bus->unplug);
+ if (bus->ctrl == NULL)
+ bus->ctrl = &rte_bus_default_ctrl;
TAILQ_INSERT_TAIL(&rte_bus_list, bus, next);
RTE_LOG(DEBUG, EAL, "Registered [%s] bus.\n", bus->name);
diff --git a/lib/librte_eal/common/include/rte_bus.h
b/lib/librte_eal/common/include/rte_bus.h
index 331d954..bd3c28e 100644
--- a/lib/librte_eal/common/include/rte_bus.h
+++ b/lib/librte_eal/common/include/rte_bus.h
@@ -183,6 +183,51 @@ struct rte_bus_conf {
enum rte_bus_probe_mode probe_mode; /**< Probe policy. */
};
+/**
+ * Bus configuration items.
+ */
+enum rte_bus_ctrl_item {
+ RTE_BUS_CTRL_PROBE_MODE = 0,
+ RTE_BUS_CTRL_ITEM_MAX,
+};
+
+/**
+ * Bus configuration operations.
+ */
+enum rte_bus_ctrl_op {
+ RTE_BUS_CTRL_GET = 0,
+ RTE_BUS_CTRL_SET,
+ RTE_BUS_CTRL_RESET,
+ RTE_BUS_CTRL_OP_MAX,
+};
+
+/**
+ * Operator for a particular rte_bus configuration item.
+ *
+ * @param arg
+ *Operation parameter.
+ *
+ * @return
+ * 0 on success
+ * !0 otherwise
+ */
+typedef int (*rte_bus_ctrl_t)(void *arg);
+
+/**
+ * Accessor to bus configuration operators.
+ *
+ * @param op
+ * Operation type.
+ *
+ * @param item
+ * Operation element.
+ *
+ * @return
+ * Operator function on success.
+ * NULL if this item is not supported.
+ */
+typedef rte_bus_ctrl_t (*rte_bus_ctrl_get_t)(enum rte_bus_ctrl_op op,
+enum rte_bus_ctrl_item item);
/**
* Get common iommu class of the all the devices on the bus. The bus may
@@ -211,6 +256,7 @@ struct rte_bus {
rte_bus_parse_t parse; /**< Parse a device name */
struct rte_bus_conf conf;/**< Bus configuration */
rte_bus_get_iommu_class_t get_iommu_class; /**< Get iommu class */
+ rte_bus_ctrl_get_t ctrl; /**< Get control operators */
};
/**
--
2.1.4
[dpdk-dev] [PATCH v1 1/8] bus: rename scan policy as probe policy
This bus configuration item is misnamed, as it actually refers to the
probing process.
Signed-off-by: Gaetan Rivet
---
drivers/bus/pci/pci_common.c | 2 +-
lib/librte_eal/common/eal_common_devargs.c | 6 +++---
lib/librte_eal/common/include/rte_bus.h| 12 ++--
3 files changed, 10 insertions(+), 10 deletions(-)
diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c
index d7a1c05..cc23a39 100644
--- a/drivers/bus/pci/pci_common.c
+++ b/drivers/bus/pci/pci_common.c
@@ -373,7 +373,7 @@ rte_pci_probe(void)
int probe_all = 0;
int ret = 0;
- if (rte_pci_bus.bus.conf.scan_mode != RTE_BUS_SCAN_WHITELIST)
+ if (rte_pci_bus.bus.conf.probe_mode != RTE_BUS_PROBE_WHITELIST)
probe_all = 1;
FOREACH_DEVICE_ON_PCIBUS(dev) {
diff --git a/lib/librte_eal/common/eal_common_devargs.c
b/lib/librte_eal/common/eal_common_devargs.c
index 6ac88d6..f5ef913 100644
--- a/lib/librte_eal/common/eal_common_devargs.c
+++ b/lib/librte_eal/common/eal_common_devargs.c
@@ -170,11 +170,11 @@ rte_eal_devargs_add(enum rte_devtype devtype, const char
*devargs_str)
bus = devargs->bus;
if (devargs->type == RTE_DEVTYPE_BLACKLISTED_PCI)
devargs->policy = RTE_DEV_BLACKLISTED;
- if (bus->conf.scan_mode == RTE_BUS_SCAN_UNDEFINED) {
+ if (bus->conf.probe_mode == RTE_BUS_PROBE_UNDEFINED) {
if (devargs->policy == RTE_DEV_WHITELISTED)
- bus->conf.scan_mode = RTE_BUS_SCAN_WHITELIST;
+ bus->conf.probe_mode = RTE_BUS_PROBE_WHITELIST;
else if (devargs->policy == RTE_DEV_BLACKLISTED)
- bus->conf.scan_mode = RTE_BUS_SCAN_BLACKLIST;
+ bus->conf.probe_mode = RTE_BUS_PROBE_BLACKLIST;
}
TAILQ_INSERT_TAIL(&devargs_list, devargs, next);
return 0;
diff --git a/lib/librte_eal/common/include/rte_bus.h
b/lib/librte_eal/common/include/rte_bus.h
index 6fb0834..331d954 100644
--- a/lib/librte_eal/common/include/rte_bus.h
+++ b/lib/librte_eal/common/include/rte_bus.h
@@ -168,19 +168,19 @@ typedef int (*rte_bus_unplug_t)(struct rte_device *dev);
typedef int (*rte_bus_parse_t)(const char *name, void *addr);
/**
- * Bus scan policies
+ * Bus probe policies
*/
-enum rte_bus_scan_mode {
- RTE_BUS_SCAN_UNDEFINED,
- RTE_BUS_SCAN_WHITELIST,
- RTE_BUS_SCAN_BLACKLIST,
+enum rte_bus_probe_mode {
+ RTE_BUS_PROBE_UNDEFINED,
+ RTE_BUS_PROBE_WHITELIST,
+ RTE_BUS_PROBE_BLACKLIST,
};
/**
* A structure used to configure bus operations.
*/
struct rte_bus_conf {
- enum rte_bus_scan_mode scan_mode; /**< Scan policy. */
+ enum rte_bus_probe_mode probe_mode; /**< Probe policy. */
};
--
2.1.4
[dpdk-dev] [PATCH v1 3/8] bus: remove probe mode configuration structure
This configuration item will be implemented within the new flexible
framework.
Signed-off-by: Gaetan Rivet
---
drivers/bus/pci/pci_common.c | 5 +
lib/librte_eal/common/eal_common_devargs.c | 8
lib/librte_eal/common/include/rte_bus.h| 8
3 files changed, 1 insertion(+), 20 deletions(-)
diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c
index cc23a39..dc69113 100644
--- a/drivers/bus/pci/pci_common.c
+++ b/drivers/bus/pci/pci_common.c
@@ -370,12 +370,9 @@ rte_pci_probe(void)
struct rte_pci_device *dev = NULL;
size_t probed = 0, failed = 0;
struct rte_devargs *devargs;
- int probe_all = 0;
+ int probe_all = 1;
int ret = 0;
- if (rte_pci_bus.bus.conf.probe_mode != RTE_BUS_PROBE_WHITELIST)
- probe_all = 1;
-
FOREACH_DEVICE_ON_PCIBUS(dev) {
probed++;
diff --git a/lib/librte_eal/common/eal_common_devargs.c
b/lib/librte_eal/common/eal_common_devargs.c
index f5ef913..e371456 100644
--- a/lib/librte_eal/common/eal_common_devargs.c
+++ b/lib/librte_eal/common/eal_common_devargs.c
@@ -156,7 +156,6 @@ int
rte_eal_devargs_add(enum rte_devtype devtype, const char *devargs_str)
{
struct rte_devargs *devargs = NULL;
- struct rte_bus *bus = NULL;
const char *dev = devargs_str;
/* use calloc instead of rte_zmalloc as it's called early at init */
@@ -167,15 +166,8 @@ rte_eal_devargs_add(enum rte_devtype devtype, const char
*devargs_str)
if (rte_eal_devargs_parse(dev, devargs))
goto fail;
devargs->type = devtype;
- bus = devargs->bus;
if (devargs->type == RTE_DEVTYPE_BLACKLISTED_PCI)
devargs->policy = RTE_DEV_BLACKLISTED;
- if (bus->conf.probe_mode == RTE_BUS_PROBE_UNDEFINED) {
- if (devargs->policy == RTE_DEV_WHITELISTED)
- bus->conf.probe_mode = RTE_BUS_PROBE_WHITELIST;
- else if (devargs->policy == RTE_DEV_BLACKLISTED)
- bus->conf.probe_mode = RTE_BUS_PROBE_BLACKLIST;
- }
TAILQ_INSERT_TAIL(&devargs_list, devargs, next);
return 0;
diff --git a/lib/librte_eal/common/include/rte_bus.h
b/lib/librte_eal/common/include/rte_bus.h
index bd3c28e..a8fb6b1 100644
--- a/lib/librte_eal/common/include/rte_bus.h
+++ b/lib/librte_eal/common/include/rte_bus.h
@@ -177,13 +177,6 @@ enum rte_bus_probe_mode {
};
/**
- * A structure used to configure bus operations.
- */
-struct rte_bus_conf {
- enum rte_bus_probe_mode probe_mode; /**< Probe policy. */
-};
-
-/**
* Bus configuration items.
*/
enum rte_bus_ctrl_item {
@@ -254,7 +247,6 @@ struct rte_bus {
rte_bus_plug_t plug; /**< Probe single device for drivers */
rte_bus_unplug_t unplug; /**< Remove single device from driver */
rte_bus_parse_t parse; /**< Parse a device name */
- struct rte_bus_conf conf;/**< Bus configuration */
rte_bus_get_iommu_class_t get_iommu_class; /**< Get iommu class */
rte_bus_ctrl_get_t ctrl; /**< Get control operators */
};
--
2.1.4
[dpdk-dev] [PATCH v1 4/8] bus: add probe mode setter
Introduce new rte_bus operation to configure the probe policy.
Implementation is required from buses interested in supporting
this configuration element.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/common/eal_common_bus.c | 24
lib/librte_eal/common/eal_common_options.c | 17 -
lib/librte_eal/common/include/rte_bus.h| 16
3 files changed, 44 insertions(+), 13 deletions(-)
diff --git a/lib/librte_eal/common/eal_common_bus.c
b/lib/librte_eal/common/eal_common_bus.c
index 65d7229..5b155c6 100644
--- a/lib/librte_eal/common/eal_common_bus.c
+++ b/lib/librte_eal/common/eal_common_bus.c
@@ -120,6 +120,30 @@ rte_bus_probe(void)
return 0;
}
+/* Configure the probing policy of a bus */
+int
+rte_bus_probe_mode_set(const char *busname,
+ enum rte_bus_probe_mode mode)
+{
+ struct rte_bus *bus;
+ rte_bus_ctrl_t probe_mode_set;
+
+ bus = rte_bus_find_by_name(busname);
+ if (bus == NULL) {
+ RTE_LOG(ERR, EAL, "Bus %s not found.\n",
+ busname);
+ return -EFAULT;
+ }
+ probe_mode_set = bus->ctrl(RTE_BUS_CTRL_SET,
+ RTE_BUS_CTRL_PROBE_MODE);
+ if (probe_mode_set == NULL) {
+ RTE_LOG(ERR, EAL, "Bus %s: probe policy cannot be
configured.\n",
+ busname);
+ return -ENOTSUP;
+ }
+ return probe_mode_set(&mode);
+}
+
/* Dump information of a single bus */
static int
bus_dump_one(FILE *f, struct rte_bus *bus)
diff --git a/lib/librte_eal/common/eal_common_options.c
b/lib/librte_eal/common/eal_common_options.c
index e40c049..630c9d2 100644
--- a/lib/librte_eal/common/eal_common_options.c
+++ b/lib/librte_eal/common/eal_common_options.c
@@ -997,29 +997,24 @@ int
eal_parse_common_option(int opt, const char *optarg,
struct internal_config *conf)
{
- static int b_used;
- static int w_used;
-
switch (opt) {
/* blacklist */
case 'b':
- if (w_used)
- goto bw_used;
+ if (rte_bus_probe_mode_set("pci", RTE_BUS_PROBE_BLACKLIST) < 0)
+ return -1;
if (eal_option_device_add(RTE_DEVTYPE_BLACKLISTED_PCI,
optarg) < 0) {
return -1;
}
- b_used = 1;
break;
/* whitelist */
case 'w':
- if (b_used)
- goto bw_used;
+ if (rte_bus_probe_mode_set("pci", RTE_BUS_PROBE_WHITELIST) < 0)
+ return -1;
if (eal_option_device_add(RTE_DEVTYPE_WHITELISTED_PCI,
optarg) < 0) {
return -1;
}
- w_used = 1;
break;
/* coremask */
case 'c':
@@ -1165,10 +1160,6 @@ eal_parse_common_option(int opt, const char *optarg,
}
return 0;
-bw_used:
- RTE_LOG(ERR, EAL, "Options blacklist (-b) and whitelist (-w) "
- "cannot be used at the same time\n");
- return -1;
}
static void
diff --git a/lib/librte_eal/common/include/rte_bus.h
b/lib/librte_eal/common/include/rte_bus.h
index a8fb6b1..93108ce 100644
--- a/lib/librte_eal/common/include/rte_bus.h
+++ b/lib/librte_eal/common/include/rte_bus.h
@@ -289,6 +289,22 @@ int rte_bus_scan(void);
int rte_bus_probe(void);
/**
+ * Configure bus probe policy.
+ *
+ * @param busname
+ * Name of the bus to configure.
+ *
+ * @param mode
+ * Configure the designated bus probe policy to this value.
+ *
+ * @return
+ * 0 on success
+ * !0 otherwise
+ */
+int rte_bus_probe_mode_set(const char *busname,
+ enum rte_bus_probe_mode mode);
+
+/**
* Dump information of all the buses registered with EAL.
*
* @param f
--
2.1.4
[dpdk-dev] [PATCH v1 5/8] bus/pci: implement ctrl operator
Add the PCI bus control operator.
This operator gives access to the probe policy setting, allowing to
read and write this configuration item. The previous existing
functionality is thus restored to the same level.
Probe policy is blacklist mode by default for the PCI bus. Configuration
is allowed once, and is then considered immutable.
Signed-off-by: Gaetan Rivet
---
drivers/bus/pci/include/rte_bus_pci.h | 1 +
drivers/bus/pci/pci_common.c | 52 ++-
2 files changed, 52 insertions(+), 1 deletion(-)
diff --git a/drivers/bus/pci/include/rte_bus_pci.h
b/drivers/bus/pci/include/rte_bus_pci.h
index e6a7998..f662705 100644
--- a/drivers/bus/pci/include/rte_bus_pci.h
+++ b/drivers/bus/pci/include/rte_bus_pci.h
@@ -155,6 +155,7 @@ struct rte_pci_driver {
*/
struct rte_pci_bus {
struct rte_bus bus; /**< Inherit the generic class */
+ enum rte_bus_probe_mode probe_mode; /**< Probe policy */
struct rte_pci_device_list device_list; /**< List of PCI devices */
struct rte_pci_driver_list driver_list; /**< List of PCI drivers */
};
diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c
index dc69113..358e232 100644
--- a/drivers/bus/pci/pci_common.c
+++ b/drivers/bus/pci/pci_common.c
@@ -370,9 +370,12 @@ rte_pci_probe(void)
struct rte_pci_device *dev = NULL;
size_t probed = 0, failed = 0;
struct rte_devargs *devargs;
- int probe_all = 1;
+ int probe_all = 0;
int ret = 0;
+ if (rte_pci_bus.probe_mode != RTE_BUS_PROBE_WHITELIST)
+ probe_all = 1;
+
FOREACH_DEVICE_ON_PCIBUS(dev) {
probed++;
@@ -515,6 +518,51 @@ pci_unplug(struct rte_device *dev)
return ret;
}
+static int
+pci_probe_mode_get(void *_mode)
+{
+ enum rte_bus_probe_mode *mode = _mode;
+
+ *mode = rte_pci_bus.probe_mode;
+ return 0;
+}
+
+static int
+pci_probe_mode_set(void *_mode)
+{
+ enum rte_bus_probe_mode *mode = _mode;
+ static int conf_done;
+
+ if (conf_done &&
+ *mode != rte_pci_bus.probe_mode) {
+ RTE_LOG(ERR, EAL, "Cannot set PCI to %s mode, bus is already
configured.\n",
+(*mode == RTE_BUS_PROBE_BLACKLIST) ?
+"blacklist" : "whitelist");
+ return -1;
+ }
+ rte_pci_bus.probe_mode = *mode;
+ conf_done = 1;
+ return 0;
+}
+
+static rte_bus_ctrl_t pci_ctrl_ops[][RTE_BUS_CTRL_OP_MAX] = {
+ [RTE_BUS_CTRL_PROBE_MODE] = {
+ [RTE_BUS_CTRL_GET] = pci_probe_mode_get,
+ [RTE_BUS_CTRL_SET] = pci_probe_mode_set,
+ },
+};
+
+static rte_bus_ctrl_t
+pci_ctrl(enum rte_bus_ctrl_op op,
+enum rte_bus_ctrl_item item)
+{
+ if (item > RTE_DIM(pci_ctrl_ops))
+ return NULL;
+ if (op > RTE_DIM(pci_ctrl_ops[item]))
+ return NULL;
+ return pci_ctrl_ops[item][op];
+}
+
struct rte_pci_bus rte_pci_bus = {
.bus = {
.scan = rte_pci_scan,
@@ -524,7 +572,9 @@ struct rte_pci_bus rte_pci_bus = {
.unplug = pci_unplug,
.parse = pci_parse,
.get_iommu_class = rte_pci_get_iommu_class,
+ .ctrl = pci_ctrl,
},
+ .probe_mode = RTE_BUS_PROBE_BLACKLIST,
.device_list = TAILQ_HEAD_INITIALIZER(rte_pci_bus.device_list),
.driver_list = TAILQ_HEAD_INITIALIZER(rte_pci_bus.driver_list),
};
--
2.1.4
[dpdk-dev] [PATCH v1 6/8] bus: add IOVA mode as a ctrl operation
Leverage the new bus control framework for the IOVA mode
configuration item.
The previous version is left for the transition in drivers
implementation and will be removed afterward.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/bsdapp/eal/rte_eal_version.map | 1 -
lib/librte_eal/common/eal_common_bus.c | 24 +++--
lib/librte_eal/common/include/rte_bus.h | 34 ++---
lib/librte_eal/linuxapp/eal/rte_eal_version.map | 1 -
4 files changed, 35 insertions(+), 25 deletions(-)
diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
index 97b3918..573869a 100644
--- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
@@ -228,7 +228,6 @@ DPDK_17.11 {
rte_eal_iova_mode;
rte_eal_mbuf_default_mempool_ops;
rte_lcore_has_role;
- rte_pci_get_iommu_class;
rte_pci_match;
} DPDK_17.08;
diff --git a/lib/librte_eal/common/eal_common_bus.c
b/lib/librte_eal/common/eal_common_bus.c
index 5b155c6..3627733 100644
--- a/lib/librte_eal/common/eal_common_bus.c
+++ b/lib/librte_eal/common/eal_common_bus.c
@@ -258,18 +258,26 @@ rte_bus_find_by_device_name(const char *str)
enum rte_iova_mode
rte_bus_get_iommu_class(void)
{
- int mode = RTE_IOVA_DC;
+ enum rte_iova_mode result;
struct rte_bus *bus;
+ result = RTE_IOVA_DC;
TAILQ_FOREACH(bus, &rte_bus_list, next) {
+ rte_bus_ctrl_t iova_mode_get;
+ enum rte_iova_mode mode;
- if (bus->get_iommu_class)
- mode |= bus->get_iommu_class();
+ iova_mode_get = bus->ctrl(RTE_BUS_CTRL_GET,
+ RTE_BUS_CTRL_IOVA_MODE);
+ if (iova_mode_get != NULL) {
+ if (iova_mode_get(&mode))
+ RTE_LOG(ERR, EAL, "Bus %s: error reading IOMMU
class\n",
+ bus->name);
+ else
+ result |= mode;
+ }
}
-
- if (mode != RTE_IOVA_VA) {
+ if (result != RTE_IOVA_VA)
/* Use default IOVA mode */
- mode = RTE_IOVA_PA;
- }
- return mode;
+ result = RTE_IOVA_PA;
+ return result;
}
diff --git a/lib/librte_eal/common/include/rte_bus.h
b/lib/librte_eal/common/include/rte_bus.h
index 93108ce..bb02d9d 100644
--- a/lib/librte_eal/common/include/rte_bus.h
+++ b/lib/librte_eal/common/include/rte_bus.h
@@ -55,21 +55,6 @@ extern "C" {
/** Double linked list of buses */
TAILQ_HEAD(rte_bus_list, rte_bus);
-
-/**
- * IOVA mapping mode.
- *
- * IOVA mapping mode is iommu programming mode of a device.
- * That device (for example: IOMMU backed DMA device) based
- * on rte_iova_mode will generate physical or virtual address.
- *
- */
-enum rte_iova_mode {
- RTE_IOVA_DC = 0,/* Don't care mode */
- RTE_IOVA_PA = (1 << 0), /* DMA using physical address */
- RTE_IOVA_VA = (1 << 1) /* DMA using virtual address */
-};
-
/**
* Bus specific scan for devices attached on the bus.
* For each bus object, the scan would be responsible for finding devices and
@@ -177,10 +162,29 @@ enum rte_bus_probe_mode {
};
/**
+ * IOVA mapping mode.
+ *
+ * IOVA mapping mode is iommu programming mode of a device.
+ * That device (for example: IOMMU backed DMA device) based
+ * on rte_iova_mode will generate physical or virtual address.
+ *
+ */
+enum rte_iova_mode {
+ RTE_IOVA_DC = 0,/* Don't care mode */
+ RTE_IOVA_PA = (1 << 0), /* DMA using physical address */
+ RTE_IOVA_VA = (1 << 1) /* DMA using virtual address */
+};
+
+/**
* Bus configuration items.
*/
enum rte_bus_ctrl_item {
RTE_BUS_CTRL_PROBE_MODE = 0,
+ /**
+* IOMMU class common to all devices on the bus.
+* If irrelevant, the bus may return RTE_IOVA_DC.
+*/
+ RTE_BUS_CTRL_IOVA_MODE,
RTE_BUS_CTRL_ITEM_MAX,
};
diff --git a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
index a8ea4ea..a2709e3 100644
--- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
@@ -232,7 +232,6 @@ DPDK_17.11 {
rte_eal_iova_mode;
rte_eal_mbuf_default_mempool_ops;
rte_lcore_has_role;
- rte_pci_get_iommu_class;
rte_pci_match;
} DPDK_17.08;
--
2.1.4
[dpdk-dev] [PATCH v1 7/8] bus/pci: implement IOVA mode getter
Implement the ctrl operator for the IOVA mode configuration item.
The previous functionality is kept identical, only the new control
framework is used.
All operators are made private as there is no reason to expose them.
Signed-off-by: Gaetan Rivet
---
drivers/bus/pci/bsd/pci.c | 9 ++---
drivers/bus/pci/include/rte_bus_pci.h | 11 ---
drivers/bus/pci/linux/pci.c | 20 +---
drivers/bus/pci/pci_common.c | 4 +++-
drivers/bus/pci/private.h | 13 +
5 files changed, 35 insertions(+), 22 deletions(-)
diff --git a/drivers/bus/pci/bsd/pci.c b/drivers/bus/pci/bsd/pci.c
index 753d914..ffb159f 100644
--- a/drivers/bus/pci/bsd/pci.c
+++ b/drivers/bus/pci/bsd/pci.c
@@ -407,11 +407,14 @@ rte_pci_scan(void)
/*
* Get iommu class of PCI devices on the bus.
*/
-enum rte_iova_mode
-rte_pci_get_iommu_class(void)
+int
+pci_iommu_class_get(void *_mode)
{
+ enum rte_iova_mode *mode = _mode;
+
/* Supports only RTE_KDRV_NIC_UIO */
- return RTE_IOVA_PA;
+ *mode = RTE_IOVA_PA;
+ return 0;
}
int
diff --git a/drivers/bus/pci/include/rte_bus_pci.h
b/drivers/bus/pci/include/rte_bus_pci.h
index f662705..a120b70 100644
--- a/drivers/bus/pci/include/rte_bus_pci.h
+++ b/drivers/bus/pci/include/rte_bus_pci.h
@@ -205,17 +205,6 @@ int
rte_pci_match(const struct rte_pci_driver *pci_drv,
const struct rte_pci_device *pci_dev);
-
-/**
- * Get iommu class of PCI devices on the bus.
- * And return their preferred iova mapping mode.
- *
- * @return
- * - enum rte_iova_mode.
- */
-enum rte_iova_mode
-rte_pci_get_iommu_class(void);
-
/**
* Map the PCI device resources in user space virtual memory address
*
diff --git a/drivers/bus/pci/linux/pci.c b/drivers/bus/pci/linux/pci.c
index 422579f..b711cdf 100644
--- a/drivers/bus/pci/linux/pci.c
+++ b/drivers/bus/pci/linux/pci.c
@@ -553,17 +553,20 @@ pci_one_device_has_iova_va(void)
/*
* Get iommu class of PCI devices on the bus.
*/
-enum rte_iova_mode
-rte_pci_get_iommu_class(void)
+int
+pci_iommu_class_get(void *_mode)
{
+ enum rte_iova_mode *mode = _mode;
bool is_bound;
bool is_vfio_noiommu_enabled = true;
bool has_iova_va;
bool is_bound_uio;
is_bound = pci_one_device_is_bound();
- if (!is_bound)
- return RTE_IOVA_DC;
+ if (!is_bound) {
+ *mode = RTE_IOVA_DC;
+ return 0;
+ }
has_iova_va = pci_one_device_has_iova_va();
is_bound_uio = pci_one_device_bound_uio();
@@ -572,8 +575,10 @@ rte_pci_get_iommu_class(void)
true : false;
#endif
- if (has_iova_va && !is_bound_uio && !is_vfio_noiommu_enabled)
- return RTE_IOVA_VA;
+ if (has_iova_va && !is_bound_uio && !is_vfio_noiommu_enabled) {
+ *mode = RTE_IOVA_VA;
+ return 0;
+ }
if (has_iova_va) {
RTE_LOG(WARNING, EAL, "Some devices want iova as va but pa will
be used because.. ");
@@ -583,7 +588,8 @@ rte_pci_get_iommu_class(void)
RTE_LOG(WARNING, EAL, "few device bound to UIO\n");
}
- return RTE_IOVA_PA;
+ *mode = RTE_IOVA_PA;
+ return 0;
}
/* Read PCI config space. */
diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c
index 358e232..bbe862b 100644
--- a/drivers/bus/pci/pci_common.c
+++ b/drivers/bus/pci/pci_common.c
@@ -550,6 +550,9 @@ static rte_bus_ctrl_t pci_ctrl_ops[][RTE_BUS_CTRL_OP_MAX] =
{
[RTE_BUS_CTRL_GET] = pci_probe_mode_get,
[RTE_BUS_CTRL_SET] = pci_probe_mode_set,
},
+ [RTE_BUS_CTRL_IOVA_MODE] = {
+ [RTE_BUS_CTRL_GET] = pci_iommu_class_get,
+ },
};
static rte_bus_ctrl_t
@@ -571,7 +574,6 @@ struct rte_pci_bus rte_pci_bus = {
.plug = pci_plug,
.unplug = pci_unplug,
.parse = pci_parse,
- .get_iommu_class = rte_pci_get_iommu_class,
.ctrl = pci_ctrl,
},
.probe_mode = RTE_BUS_PROBE_BLACKLIST,
diff --git a/drivers/bus/pci/private.h b/drivers/bus/pci/private.h
index fdc2c81..ee13855 100644
--- a/drivers/bus/pci/private.h
+++ b/drivers/bus/pci/private.h
@@ -171,4 +171,17 @@ void pci_uio_free_resource(struct rte_pci_device *dev,
int pci_uio_map_resource_by_index(struct rte_pci_device *dev, int res_idx,
struct mapped_pci_resource *uio_res, int map_idx);
+/**
+ * Get iommu class of PCI devices on the bus.
+ * Return their preferred iova mapping mode.
+ *
+ * @param _mode
+ * Generic address to an (enum rte_iova_mode)
+ * @return
+ * 0 on success
+ * !0 otherwise
+ */
+int
+pci_iommu_class_get(void *_mode);
+
#endif /* _PCI_PRIVATE_H_ */
--
2.1.4
[dpdk-dev] [PATCH v2 01/18] eal: prepend busname on legacy device declaration
Legacy device options (-b, -w, --vdev) need to prepend their bus name to
user parameters for backward compatibility.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/common/eal_common_options.c | 17 -
1 file changed, 12 insertions(+), 5 deletions(-)
diff --git a/lib/librte_eal/common/eal_common_options.c
b/lib/librte_eal/common/eal_common_options.c
index 630c9d2..d57cb5d 100644
--- a/lib/librte_eal/common/eal_common_options.c
+++ b/lib/librte_eal/common/eal_common_options.c
@@ -143,13 +143,16 @@ static int mem_parsed;
static int core_parsed;
static int
-eal_option_device_add(enum rte_devtype type, const char *optarg)
+eal_option_device_add(enum rte_devtype type,
+ const char *busname, const char *optarg)
{
struct device_option *devopt;
size_t optlen;
int ret;
optlen = strlen(optarg) + 1;
+ if (busname != NULL)
+ optlen += strlen(optarg) + 1;
devopt = calloc(1, sizeof(*devopt) + optlen);
if (devopt == NULL) {
RTE_LOG(ERR, EAL, "Unable to allocate device option\n");
@@ -157,7 +160,11 @@ eal_option_device_add(enum rte_devtype type, const char
*optarg)
}
devopt->type = type;
- ret = snprintf(devopt->arg, optlen, "%s", optarg);
+ if (busname != NULL)
+ ret = snprintf(devopt->arg, optlen, "%s:%s",
+ busname, optarg);
+ else
+ ret = snprintf(devopt->arg, optlen, "%s", optarg);
if (ret < 0) {
RTE_LOG(ERR, EAL, "Unable to copy device option\n");
free(devopt);
@@ -1003,7 +1010,7 @@ eal_parse_common_option(int opt, const char *optarg,
if (rte_bus_probe_mode_set("pci", RTE_BUS_PROBE_BLACKLIST) < 0)
return -1;
if (eal_option_device_add(RTE_DEVTYPE_BLACKLISTED_PCI,
- optarg) < 0) {
+ "pci", optarg) < 0) {
return -1;
}
break;
@@ -1012,7 +1019,7 @@ eal_parse_common_option(int opt, const char *optarg,
if (rte_bus_probe_mode_set("pci", RTE_BUS_PROBE_WHITELIST) < 0)
return -1;
if (eal_option_device_add(RTE_DEVTYPE_WHITELISTED_PCI,
- optarg) < 0) {
+ "pci", optarg) < 0) {
return -1;
}
break;
@@ -1122,7 +1129,7 @@ eal_parse_common_option(int opt, const char *optarg,
case OPT_VDEV_NUM:
if (eal_option_device_add(RTE_DEVTYPE_VIRTUAL,
- optarg) < 0) {
+ "vdev", optarg) < 0) {
return -1;
}
break;
--
2.1.4
[dpdk-dev] [PATCH v1 8/8] bus: remove redundant IOVA mode getter
This configuration element is now accessible through the bus control
framework and can be removed from the generic bus structure.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/common/include/rte_bus.h | 33 +
1 file changed, 9 insertions(+), 24 deletions(-)
diff --git a/lib/librte_eal/common/include/rte_bus.h
b/lib/librte_eal/common/include/rte_bus.h
index bb02d9d..1cae96e 100644
--- a/lib/librte_eal/common/include/rte_bus.h
+++ b/lib/librte_eal/common/include/rte_bus.h
@@ -227,19 +227,6 @@ typedef rte_bus_ctrl_t (*rte_bus_ctrl_get_t)(enum
rte_bus_ctrl_op op,
enum rte_bus_ctrl_item item);
/**
- * Get common iommu class of the all the devices on the bus. The bus may
- * check that those devices are attached to iommu driver.
- * If no devices are attached to the bus. The bus may return with don't care
- * (_DC) value.
- * Otherwise, The bus will return appropriate _pa or _va iova mode.
- *
- * @return
- * enum rte_iova_mode value.
- */
-typedef enum rte_iova_mode (*rte_bus_get_iommu_class_t)(void);
-
-
-/**
* A structure describing a generic bus.
*/
struct rte_bus {
@@ -251,7 +238,6 @@ struct rte_bus {
rte_bus_plug_t plug; /**< Probe single device for drivers */
rte_bus_unplug_t unplug; /**< Remove single device from driver */
rte_bus_parse_t parse; /**< Parse a device name */
- rte_bus_get_iommu_class_t get_iommu_class; /**< Get iommu class */
rte_bus_ctrl_get_t ctrl; /**< Get control operators */
};
@@ -309,6 +295,15 @@ int rte_bus_probe_mode_set(const char *busname,
enum rte_bus_probe_mode mode);
/**
+ * Get the common iommu class of devices bound on to buses available in the
+ * system. The default mode is PA.
+ *
+ * @return
+ * enum rte_iova_mode value.
+ */
+enum rte_iova_mode rte_bus_get_iommu_class(void);
+
+/**
* Dump information of all the buses registered with EAL.
*
* @param f
@@ -368,16 +363,6 @@ struct rte_bus *rte_bus_find_by_device(const struct
rte_device *dev);
*/
struct rte_bus *rte_bus_find_by_name(const char *busname);
-
-/**
- * Get the common iommu class of devices bound on to buses available in the
- * system. The default mode is PA.
- *
- * @return
- * enum rte_iova_mode value.
- */
-enum rte_iova_mode rte_bus_get_iommu_class(void);
-
/**
* Helper for Bus registration.
* The constructor has higher priority than PMD constructors.
--
2.1.4
[dpdk-dev] [PATCH v2 00/18] devargs cleanup
The use of rte_devargs is inconsistent in the light of new functionalities such as device hotplug. Most of its API is still experimental and needs stabilization. Older functions were deprecated and need to be rewritten or removed. The rte_devtype is meant to disappear. v2: Big rework. * Enact requiring bus name prepended in rte_devargs parsing functions. * Remove rte_devtype. Use new probe mode setter along with generic bus reference within rte_devargs. * Rework devargs parsing API. The function is now variadic, does not enforce bus rules on the devargs being inserted as the bus has been configured previously. Old parsing function is removed. * Expose bus guessing from device name. This uses the "parse" bus operator, which may be meant to disappear. This is optional, but nice to have in a transition period. * Introduce new --dev generic device declaration parameter. This patchset depends on: Move PCI away from the EAL http://dpdk.org/ml/archives/dev/2017-August/073512.html Bus control framework http://dpdk.org/ml/archives/dev/2017-October/078752.html Gaetan Rivet (18): eal: prepend busname on legacy device declaration eal: remove generic devtype devargs: introduce iterator devargs: introduce foreach macro vdev: do not reference devargs list bus/pci: do not reference devargs list test: remove devargs unit tests devargs: make devargs list private devargs: make parsing variadic devargs: require bus name prefix devargs: simplify implementation eal: add generic device declaration parameter bus: make device recognition function public net/failsafe: keep legacy sub-device declaration ether: use new devargs parsing function devargs: remove old devargs parsing function devargs: use proper prefix doc: remove devargs deprecation notices MAINTAINERS | 1 - app/test-pmd/cmdline.c | 2 +- doc/guides/rel_notes/deprecation.rst| 13 --- drivers/bus/pci/pci_common.c| 22 +--- drivers/net/failsafe/failsafe_args.c| 11 +- examples/bond/main.c| 2 +- lib/librte_eal/bsdapp/eal/rte_eal_version.map | 15 ++- lib/librte_eal/common/eal_common_dev.c | 39 ++- lib/librte_eal/common/eal_common_devargs.c | 129 +-- lib/librte_eal/common/eal_common_options.c | 47 ++--- lib/librte_eal/common/eal_common_vdev.c | 11 +- lib/librte_eal/common/eal_options.h | 2 + lib/librte_eal/common/include/rte_bus.h | 12 +++ lib/librte_eal/common/include/rte_dev.h | 8 -- lib/librte_eal/common/include/rte_devargs.h | 120 -- lib/librte_eal/linuxapp/eal/rte_eal_version.map | 15 ++- lib/librte_ether/rte_ethdev.c | 11 +- test/test/Makefile | 1 - test/test/commands.c| 2 +- test/test/test_devargs.c| 131 20 files changed, 186 insertions(+), 408 deletions(-) delete mode 100644 test/test/test_devargs.c -- 2.1.4
[dpdk-dev] [PATCH v2 02/18] eal: remove generic devtype
The devtype is now entirely defined by the device bus. As such, it is
already characterized by the bus identifier within an rte_devargs.
The rte_devtype enum can disappear, along with crutches added during
this transition.
rte_eal_devargs_type_count becomes useless and is removed.
Signed-off-by: Gaetan Rivet
---
drivers/bus/pci/pci_common.c| 16 ++
lib/librte_eal/bsdapp/eal/rte_eal_version.map | 1 -
lib/librte_eal/common/eal_common_devargs.c | 20 +
lib/librte_eal/common/eal_common_options.c | 19 +---
lib/librte_eal/common/include/rte_dev.h | 8 ---
lib/librte_eal/common/include/rte_devargs.h | 29 +
lib/librte_eal/linuxapp/eal/rte_eal_version.map | 1 -
7 files changed, 9 insertions(+), 85 deletions(-)
diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c
index bbe862b..5fbcf11 100644
--- a/drivers/bus/pci/pci_common.c
+++ b/drivers/bus/pci/pci_common.c
@@ -172,15 +172,6 @@ rte_pci_probe_one_driver(struct rte_pci_driver *dr,
loc->domain, loc->bus, loc->devid, loc->function,
dev->device.numa_node);
- /* no initialization when blacklisted, return without error */
- if (dev->device.devargs != NULL &&
- dev->device.devargs->policy ==
- RTE_DEV_BLACKLISTED) {
- RTE_LOG(INFO, EAL, " Device is blacklisted, not"
- " initializing\n");
- return 1;
- }
-
if (dev->device.numa_node < 0) {
RTE_LOG(WARNING, EAL, " Invalid NUMA socket, default to 0\n");
dev->device.numa_node = 0;
@@ -380,11 +371,8 @@ rte_pci_probe(void)
probed++;
devargs = dev->device.devargs;
- /* probe all or only whitelisted devices */
- if (probe_all)
- ret = pci_probe_all_drivers(dev);
- else if (devargs != NULL &&
- devargs->policy == RTE_DEV_WHITELISTED)
+ /* probe all or only declared devices */
+ if (probe_all ^ (devargs != NULL))
ret = pci_probe_all_drivers(dev);
if (ret < 0) {
RTE_LOG(ERR, EAL, "Requested device " PCI_PRI_FMT
diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
index 573869a..47416a5 100644
--- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
@@ -22,7 +22,6 @@ DPDK_2.0 {
rte_eal_alarm_set;
rte_eal_devargs_add;
rte_eal_devargs_dump;
- rte_eal_devargs_type_count;
rte_eal_get_configuration;
rte_eal_get_lcore_state;
rte_eal_get_physmem_layout;
diff --git a/lib/librte_eal/common/eal_common_devargs.c
b/lib/librte_eal/common/eal_common_devargs.c
index e371456..2fddbfa 100644
--- a/lib/librte_eal/common/eal_common_devargs.c
+++ b/lib/librte_eal/common/eal_common_devargs.c
@@ -153,7 +153,7 @@ rte_eal_devargs_insert(struct rte_devargs *da)
/* store a whitelist parameter for later parsing */
int
-rte_eal_devargs_add(enum rte_devtype devtype, const char *devargs_str)
+rte_eal_devargs_add(const char *devargs_str)
{
struct rte_devargs *devargs = NULL;
const char *dev = devargs_str;
@@ -165,9 +165,6 @@ rte_eal_devargs_add(enum rte_devtype devtype, const char
*devargs_str)
if (rte_eal_devargs_parse(dev, devargs))
goto fail;
- devargs->type = devtype;
- if (devargs->type == RTE_DEVTYPE_BLACKLISTED_PCI)
- devargs->policy = RTE_DEV_BLACKLISTED;
TAILQ_INSERT_TAIL(&devargs_list, devargs, next);
return 0;
@@ -198,21 +195,6 @@ rte_eal_devargs_remove(const char *busname, const char
*devname)
return 1;
}
-/* count the number of devices of a specified type */
-unsigned int
-rte_eal_devargs_type_count(enum rte_devtype devtype)
-{
- struct rte_devargs *devargs;
- unsigned int count = 0;
-
- TAILQ_FOREACH(devargs, &devargs_list, next) {
- if (devargs->type != devtype)
- continue;
- count++;
- }
- return count;
-}
-
/* dump the user devices on the console */
void
rte_eal_devargs_dump(FILE *f)
diff --git a/lib/librte_eal/common/eal_common_options.c
b/lib/librte_eal/common/eal_common_options.c
index d57cb5d..603df27 100644
--- a/lib/librte_eal/common/eal_common_options.c
+++ b/lib/librte_eal/common/eal_common_options.c
@@ -131,7 +131,6 @@ TAILQ_HEAD(device_option_list, device_option);
struct device_option {
TAILQ_ENTRY(device_option) next;
- enum rte_devtype type;
char arg[];
};
@@ -143,8 +142,7 @@ static int mem_parsed;
static int core_parsed;
static int
-eal_option_device_add(enum rte_devtype type,
- const char *busname, co
[dpdk-dev] [PATCH v2 03/18] devargs: introduce iterator
In preparation to making devargs_list private.
Bus drivers generally need to access rte_devargs pertaining to their
operations. This match is a common operation for bus drivers.
Add a new accessor for the rte_devargs list.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/bsdapp/eal/rte_eal_version.map | 1 +
lib/librte_eal/common/eal_common_devargs.c | 19 +++
lib/librte_eal/common/include/rte_devargs.h | 18 ++
lib/librte_eal/linuxapp/eal/rte_eal_version.map | 1 +
4 files changed, 39 insertions(+)
diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
index 47416a5..01ae0c7 100644
--- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
@@ -189,6 +189,7 @@ EXPERIMENTAL {
global:
rte_eal_devargs_insert;
+ rte_eal_devargs_next;
rte_eal_devargs_parse;
rte_eal_devargs_remove;
rte_eal_hotplug_add;
diff --git a/lib/librte_eal/common/eal_common_devargs.c
b/lib/librte_eal/common/eal_common_devargs.c
index 2fddbfa..614f1c5 100644
--- a/lib/librte_eal/common/eal_common_devargs.c
+++ b/lib/librte_eal/common/eal_common_devargs.c
@@ -208,3 +208,22 @@ rte_eal_devargs_dump(FILE *f)
devargs->name, devargs->args);
}
}
+
+/* bus-aware rte_devargs iterator. */
+struct rte_devargs *
+rte_eal_devargs_next(const char *busname, const struct rte_devargs *start)
+{
+ struct rte_devargs *da;
+
+ if (start != NULL)
+ da = TAILQ_NEXT(start, next);
+ else
+ da = TAILQ_FIRST(&devargs_list);
+ while (da != NULL) {
+ if (busname == NULL ||
+ (strcmp(busname, da->bus->name) == 0))
+ return da;
+ da = TAILQ_NEXT(da, next);
+ }
+ return NULL;
+}
diff --git a/lib/librte_eal/common/include/rte_devargs.h
b/lib/librte_eal/common/include/rte_devargs.h
index e50c166..0eec406 100644
--- a/lib/librte_eal/common/include/rte_devargs.h
+++ b/lib/librte_eal/common/include/rte_devargs.h
@@ -188,6 +188,24 @@ int rte_eal_devargs_remove(const char *busname, const char
*devname);
*/
void rte_eal_devargs_dump(FILE *f);
+/**
+ * Find next rte_devargs matching the provided bus name.
+ *
+ * @param busname
+ * Limit the iteration to bus matching this name.
+ * Will return any next rte_devargs if NULL.
+ *
+ * @param start
+ * Starting iteration point. The iteration will start at
+ * the first rte_devargs if NULL.
+ *
+ * @return
+ * Next rte_devargs entry matching the requested bus,
+ * NULL if there is none.
+ */
+struct rte_devargs *
+rte_eal_devargs_next(const char *busname, const struct rte_devargs *start);
+
#ifdef __cplusplus
}
#endif
diff --git a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
index e1e2a50..576de56 100644
--- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
@@ -193,6 +193,7 @@ EXPERIMENTAL {
global:
rte_eal_devargs_insert;
+ rte_eal_devargs_next;
rte_eal_devargs_parse;
rte_eal_devargs_remove;
rte_eal_hotplug_add;
--
2.1.4
[dpdk-dev] [PATCH v2 05/18] vdev: do not reference devargs list
This list should not be operated upon by drivers.
Use the public API to achieve the same functionalities.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/common/eal_common_vdev.c | 11 +++
1 file changed, 3 insertions(+), 8 deletions(-)
diff --git a/lib/librte_eal/common/eal_common_vdev.c
b/lib/librte_eal/common/eal_common_vdev.c
index f7e547a..a7410a6 100644
--- a/lib/librte_eal/common/eal_common_vdev.c
+++ b/lib/librte_eal/common/eal_common_vdev.c
@@ -192,7 +192,7 @@ rte_vdev_init(const char *name, const char *args)
goto fail;
}
- TAILQ_INSERT_TAIL(&devargs_list, devargs, next);
+ rte_eal_devargs_insert(devargs);
TAILQ_INSERT_TAIL(&vdev_device_list, dev, next);
return 0;
@@ -242,10 +242,8 @@ rte_vdev_uninit(const char *name)
TAILQ_REMOVE(&vdev_device_list, dev, next);
- TAILQ_REMOVE(&devargs_list, devargs, next);
+ rte_eal_devargs_remove(devargs->bus->name, devargs->name);
- free(devargs->args);
- free(devargs);
free(dev);
return 0;
}
@@ -257,10 +255,7 @@ vdev_scan(void)
struct rte_devargs *devargs;
/* for virtual devices we scan the devargs_list populated via cmdline */
- TAILQ_FOREACH(devargs, &devargs_list, next) {
-
- if (devargs->bus != &rte_vdev_bus)
- continue;
+ RTE_EAL_DEVARGS_FOREACH("vdev", devargs) {
dev = find_vdev(devargs->name);
if (dev)
--
2.1.4
[dpdk-dev] [PATCH v2 04/18] devargs: introduce foreach macro
Introduce new rte_devargs accessor allowing to iterate over all rte_devargs pertaining to a bus. Signed-off-by: Gaetan Rivet --- lib/librte_eal/common/include/rte_devargs.h | 8 1 file changed, 8 insertions(+) diff --git a/lib/librte_eal/common/include/rte_devargs.h b/lib/librte_eal/common/include/rte_devargs.h index 0eec406..6222677 100644 --- a/lib/librte_eal/common/include/rte_devargs.h +++ b/lib/librte_eal/common/include/rte_devargs.h @@ -206,6 +206,14 @@ void rte_eal_devargs_dump(FILE *f); struct rte_devargs * rte_eal_devargs_next(const char *busname, const struct rte_devargs *start); +/** + * Iterate over all rte_devargs for a specific bus. + */ +#define RTE_EAL_DEVARGS_FOREACH(busname, da) \ + for (da = rte_eal_devargs_next(busname, NULL); \ +da != NULL; \ +da = rte_eal_devargs_next(busname, da)) \ + #ifdef __cplusplus } #endif -- 2.1.4
[dpdk-dev] [PATCH v2 07/18] test: remove devargs unit tests
The current test will not be compatible anymore with a private
devargs list.
Moreover, the new functions should have new tests, while the existing
API will be removed.
The current unit tests are thus obsolete and hereby removed.
Signed-off-by: Gaetan Rivet
---
MAINTAINERS | 1 -
test/test/Makefile | 1 -
test/test/test_devargs.c | 131 ---
3 files changed, 133 deletions(-)
delete mode 100644 test/test/test_devargs.c
diff --git a/MAINTAINERS b/MAINTAINERS
index b8b5441..6c174ef 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -96,7 +96,6 @@ F: test/test/test_common.c
F: test/test/test_cpuflags.c
F: test/test/test_cycles.c
F: test/test/test_debug.c
-F: test/test/test_devargs.c
F: test/test/test_eal*
F: test/test/test_errno.c
F: test/test/test_interrupts.c
diff --git a/test/test/Makefile b/test/test/Makefile
index 61e4699..3d76e5e 100644
--- a/test/test/Makefile
+++ b/test/test/Makefile
@@ -184,7 +184,6 @@ SRCS-$(CONFIG_RTE_LIBRTE_DISTRIBUTOR) +=
test_distributor_perf.c
SRCS-$(CONFIG_RTE_LIBRTE_REORDER) += test_reorder.c
-SRCS-y += test_devargs.c
SRCS-y += virtual_pmd.c
SRCS-y += packet_burst_generator.c
SRCS-$(CONFIG_RTE_LIBRTE_ACL) += test_acl.c
diff --git a/test/test/test_devargs.c b/test/test/test_devargs.c
deleted file mode 100644
index 18f54ed..000
--- a/test/test/test_devargs.c
+++ /dev/null
@@ -1,131 +0,0 @@
-/*-
- * BSD LICENSE
- *
- * Copyright 2014 6WIND S.A.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- *
- * * Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- * * Redistributions in binary form must reproduce the above copyright
- * notice, this list of conditions and the following disclaimer in
- * the documentation and/or other materials provided with the
- * distribution.
- * * Neither the name of 6WIND S.A nor the names of its contributors
- * may be used to endorse or promote products derived from this
- * software without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
- * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
- * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
- * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
- * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
- * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
- * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
- * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
- * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
- * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- */
-
-#include
-#include
-#include
-#include
-
-#include
-#include
-
-#include "test.h"
-
-/* clear devargs list that was modified by the test */
-static void free_devargs_list(void)
-{
- struct rte_devargs *devargs;
-
- while (!TAILQ_EMPTY(&devargs_list)) {
- devargs = TAILQ_FIRST(&devargs_list);
- TAILQ_REMOVE(&devargs_list, devargs, next);
- free(devargs->args);
- free(devargs);
- }
-}
-
-static int
-test_devargs(void)
-{
- struct rte_devargs_list save_devargs_list;
- struct rte_devargs *devargs;
-
- /* save the real devargs_list, it is restored at the end of the test */
- save_devargs_list = devargs_list;
- TAILQ_INIT(&devargs_list);
-
- /* test valid cases */
- if (rte_eal_devargs_add(RTE_DEVTYPE_WHITELISTED_PCI, "08:00.1") < 0)
- goto fail;
- if (rte_eal_devargs_add(RTE_DEVTYPE_WHITELISTED_PCI, ":5:00.0") < 0)
- goto fail;
- if (rte_eal_devargs_add(RTE_DEVTYPE_BLACKLISTED_PCI, "04:00.0,arg=val")
< 0)
- goto fail;
- if (rte_eal_devargs_add(RTE_DEVTYPE_BLACKLISTED_PCI, ":01:00.1") <
0)
- goto fail;
- if (rte_eal_devargs_type_count(RTE_DEVTYPE_WHITELISTED_PCI) != 2)
- goto fail;
- if (rte_eal_devargs_type_count(RTE_DEVTYPE_BLACKLISTED_PCI) != 2)
- goto fail;
- if (rte_eal_devargs_type_count(RTE_DEVTYPE_VIRTUAL) != 0)
- goto fail;
- if (rte_eal_devargs_add(RTE_DEVTYPE_VIRTUAL, "net_ring0") < 0)
- goto fail;
- if (rte_eal_devargs_add(RTE_DEVTYPE_VIRTUAL,
"net_ring1,key=val,k2=val2") < 0)
- goto fail;
- if (rte_eal_devargs_type_count(RTE_DEVTYPE_VIRTUAL) != 2)
- goto fail;
- free_devargs_list();
-
- /* check virtual device with argument parsing */
- if
[dpdk-dev] [PATCH v2 06/18] bus/pci: do not reference devargs list
This list should not be used by drivers.
Use the public API instead.
Signed-off-by: Gaetan Rivet
---
drivers/bus/pci/pci_common.c | 6 +-
1 file changed, 1 insertion(+), 5 deletions(-)
diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c
index 5fbcf11..0b64d20 100644
--- a/drivers/bus/pci/pci_common.c
+++ b/drivers/bus/pci/pci_common.c
@@ -75,12 +75,8 @@ static struct rte_devargs *pci_devargs_lookup(struct
rte_pci_device *dev)
{
struct rte_devargs *devargs;
struct rte_pci_addr addr;
- struct rte_bus *pbus;
- pbus = rte_bus_find_by_name("pci");
- TAILQ_FOREACH(devargs, &devargs_list, next) {
- if (devargs->bus != pbus)
- continue;
+ RTE_EAL_DEVARGS_FOREACH("pci", devargs) {
devargs->bus->parse(devargs->name, &addr);
if (!rte_eal_compare_pci_addr(&dev->addr, &addr))
return devargs;
--
2.1.4
[dpdk-dev] [PATCH v2 09/18] devargs: make parsing variadic
rte_eal_devargs_parse can be used by EAL subsystems, drivers,
applications alike.
Device parameters may be presented with different structure each time;
as a single declaration string or several strings each describing
different parts of the declaration.
To simplify the use of this parsing facility, its parameters are made
variadic.
Signed-off-by: Gaetan Rivet
---
drivers/net/failsafe/failsafe_args.c| 2 +-
lib/librte_eal/common/eal_common_dev.c | 33 -
lib/librte_eal/common/eal_common_devargs.c | 15 ++---
lib/librte_eal/common/include/rte_devargs.h | 25 +++---
4 files changed, 30 insertions(+), 45 deletions(-)
diff --git a/drivers/net/failsafe/failsafe_args.c
b/drivers/net/failsafe/failsafe_args.c
index cfc83e3..08ce4ad 100644
--- a/drivers/net/failsafe/failsafe_args.c
+++ b/drivers/net/failsafe/failsafe_args.c
@@ -88,7 +88,7 @@ fs_parse_device(struct sub_device *sdev, char *args)
d = &sdev->devargs;
DEBUG("%s", args);
- ret = rte_eal_devargs_parse(args, d);
+ ret = rte_eal_devargs_parse(d, "%s", args);
if (ret) {
DEBUG("devargs parsing failed with code %d", ret);
return ret;
diff --git a/lib/librte_eal/common/eal_common_dev.c
b/lib/librte_eal/common/eal_common_dev.c
index e251275..b965e56 100644
--- a/lib/librte_eal/common/eal_common_dev.c
+++ b/lib/librte_eal/common/eal_common_dev.c
@@ -127,29 +127,12 @@ int rte_eal_dev_detach(struct rte_device *dev)
return ret;
}
-static char *
-full_dev_name(const char *bus, const char *dev, const char *args)
-{
- char *name;
- size_t len;
-
- len = snprintf(NULL, 0, "%s:%s,%s", bus, dev, args) + 1;
- name = calloc(1, len);
- if (name == NULL) {
- RTE_LOG(ERR, EAL, "Could not allocate full device name\n");
- return NULL;
- }
- snprintf(name, len, "%s:%s,%s", bus, dev, args);
- return name;
-}
-
int rte_eal_hotplug_add(const char *busname, const char *devname,
const char *devargs)
{
struct rte_bus *bus;
struct rte_device *dev;
struct rte_devargs *da;
- char *name;
int ret;
bus = rte_bus_find_by_name(busname);
@@ -164,17 +147,12 @@ int rte_eal_hotplug_add(const char *busname, const char
*devname,
return -ENOTSUP;
}
- name = full_dev_name(busname, devname, devargs);
- if (name == NULL)
+ da = calloc(1, sizeof(*da));
+ if (da == NULL)
return -ENOMEM;
- da = calloc(1, sizeof(*da));
- if (da == NULL) {
- ret = -ENOMEM;
- goto err_name;
- }
-
- ret = rte_eal_devargs_parse(name, da);
+ ret = rte_eal_devargs_parse(da, "%s:%s,%s",
+ busname, devname, devargs);
if (ret)
goto err_devarg;
@@ -200,7 +178,6 @@ int rte_eal_hotplug_add(const char *busname, const char
*devname,
dev->name);
goto err_devarg;
}
- free(name);
return 0;
err_devarg:
@@ -208,8 +185,6 @@ int rte_eal_hotplug_add(const char *busname, const char
*devname,
free(da->args);
free(da);
}
-err_name:
- free(name);
return ret;
}
diff --git a/lib/librte_eal/common/eal_common_devargs.c
b/lib/librte_eal/common/eal_common_devargs.c
index 0f81f22..a21cc1a 100644
--- a/lib/librte_eal/common/eal_common_devargs.c
+++ b/lib/librte_eal/common/eal_common_devargs.c
@@ -39,6 +39,7 @@
#include
#include
+#include
#include
#include
@@ -89,15 +90,23 @@ bus_name_cmp(const struct rte_bus *bus, const void *name)
}
int
-rte_eal_devargs_parse(const char *dev, struct rte_devargs *da)
+rte_eal_devargs_parse(struct rte_devargs *da, const char *format, ...)
{
struct rte_bus *bus = NULL;
+ va_list ap;
+ va_start(ap, format);
+ char dev[vsnprintf(NULL, 0, format, ap) + 1];
const char *devname;
const size_t maxlen = sizeof(da->name);
size_t i;
- if (dev == NULL || da == NULL)
+ va_end(ap);
+ if (da == NULL)
return -EINVAL;
+
+ va_start(ap, format);
+ vsnprintf(dev, sizeof(dev), format, ap);
+ va_end(ap);
/* Retrieve eventual bus info */
do {
devname = dev;
@@ -166,7 +175,7 @@ rte_eal_devargs_add(const char *devargs_str)
if (devargs == NULL)
goto fail;
- if (rte_eal_devargs_parse(dev, devargs))
+ if (rte_eal_devargs_parse(devargs, "%s", dev))
goto fail;
TAILQ_INSERT_TAIL(&devargs_list, devargs, next);
return 0;
diff --git a/lib/librte_eal/common/include/rte_devargs.h
b/lib/librte_eal/common/include/rte_devargs.h
index 5f4ad33..1fe03d6 100644
--- a/lib/librte_eal/common/include/rte_devargs.h
+++ b/lib/librte_eal/common/inc
[dpdk-dev] [PATCH v2 08/18] devargs: make devargs list private
Initially, rte_devargs was meant to be populated once and sometimes
accessed, then never emptied.
With the new hotplug functionality having better standing, new usage
appeared with repeated addition of devices and their subsequent removal.
Exposing devargs_list pushed bus drivers and libraries to be careless
and inconsistent in their memory management. Making it private will
allow to rationalize this part of the EAL and ensure that fewer memory
leaks occur during operations.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/bsdapp/eal/rte_eal_version.map | 1 -
lib/librte_eal/common/eal_common_devargs.c | 3 +++
lib/librte_eal/common/include/rte_devargs.h | 6 --
lib/librte_eal/linuxapp/eal/rte_eal_version.map | 1 -
4 files changed, 3 insertions(+), 8 deletions(-)
diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
index 01ae0c7..0d693c8 100644
--- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
@@ -2,7 +2,6 @@ DPDK_2.0 {
global:
__rte_panic;
- devargs_list;
eal_parse_sysfs_value;
eal_timer_source;
lcore_config;
diff --git a/lib/librte_eal/common/eal_common_devargs.c
b/lib/librte_eal/common/eal_common_devargs.c
index 614f1c5..0f81f22 100644
--- a/lib/librte_eal/common/eal_common_devargs.c
+++ b/lib/librte_eal/common/eal_common_devargs.c
@@ -45,6 +45,9 @@
#include
#include "eal_private.h"
+/** user device double-linked queue type definition */
+TAILQ_HEAD(rte_devargs_list, rte_devargs);
+
/** Global list of user devices */
struct rte_devargs_list devargs_list =
TAILQ_HEAD_INITIALIZER(devargs_list);
diff --git a/lib/librte_eal/common/include/rte_devargs.h
b/lib/librte_eal/common/include/rte_devargs.h
index 6222677..5f4ad33 100644
--- a/lib/librte_eal/common/include/rte_devargs.h
+++ b/lib/librte_eal/common/include/rte_devargs.h
@@ -73,12 +73,6 @@ struct rte_devargs {
char *args;
};
-/** user device double-linked queue type definition */
-TAILQ_HEAD(rte_devargs_list, rte_devargs);
-
-/** Global list of user devices */
-extern struct rte_devargs_list devargs_list;
-
/**
* Parse a devargs string.
*
diff --git a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
index 576de56..9c0251e 100644
--- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
@@ -2,7 +2,6 @@ DPDK_2.0 {
global:
__rte_panic;
- devargs_list;
eal_parse_sysfs_value;
eal_timer_source;
lcore_config;
--
2.1.4
[dpdk-dev] [PATCH v2 10/18] devargs: require bus name prefix
The EAL now requires the bus to be prepended to the device declaration
string.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/common/eal_common_devargs.c | 28 +---
1 file changed, 9 insertions(+), 19 deletions(-)
diff --git a/lib/librte_eal/common/eal_common_devargs.c
b/lib/librte_eal/common/eal_common_devargs.c
index a21cc1a..49cc3b8 100644
--- a/lib/librte_eal/common/eal_common_devargs.c
+++ b/lib/librte_eal/common/eal_common_devargs.c
@@ -107,18 +107,17 @@ rte_eal_devargs_parse(struct rte_devargs *da, const char
*format, ...)
va_start(ap, format);
vsnprintf(dev, sizeof(dev), format, ap);
va_end(ap);
- /* Retrieve eventual bus info */
- do {
- devname = dev;
- bus = rte_bus_find(bus, bus_name_cmp, dev);
- if (bus == NULL)
- break;
- devname = dev + strlen(bus->name) + 1;
- if (rte_bus_find_by_device_name(devname) == bus)
- break;
- } while (1);
+ /* Retrieve bus info */
+ bus = rte_bus_find(bus, bus_name_cmp, dev);
+ if (bus == NULL) {
+ fprintf(stderr, "ERROR: failed to parse bus from \"%s\"\n",
+ dev);
+ return -EFAULT;
+ }
+ da->bus = bus;
/* Store device name */
i = 0;
+ devname = dev + strlen(bus->name) + 1;
while (devname[i] != '\0' && devname[i] != ',') {
da->name[i] = devname[i];
i++;
@@ -130,15 +129,6 @@ rte_eal_devargs_parse(struct rte_devargs *da, const char
*format, ...)
}
}
da->name[i] = '\0';
- if (bus == NULL) {
- bus = rte_bus_find_by_device_name(da->name);
- if (bus == NULL) {
- fprintf(stderr, "ERROR: failed to parse device
\"%s\"\n",
- da->name);
- return -EFAULT;
- }
- }
- da->bus = bus;
/* Parse eventual device arguments */
if (devname[i] == ',')
da->args = strdup(&devname[i + 1]);
--
2.1.4
[dpdk-dev] [PATCH v2 11/18] devargs: simplify implementation
Re-use existing code, remove incorrect comments.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/common/eal_common_devargs.c | 8 +++-
1 file changed, 3 insertions(+), 5 deletions(-)
diff --git a/lib/librte_eal/common/eal_common_devargs.c
b/lib/librte_eal/common/eal_common_devargs.c
index 49cc3b8..1d87cd9 100644
--- a/lib/librte_eal/common/eal_common_devargs.c
+++ b/lib/librte_eal/common/eal_common_devargs.c
@@ -153,21 +153,19 @@ rte_eal_devargs_insert(struct rte_devargs *da)
return 0;
}
-/* store a whitelist parameter for later parsing */
int
-rte_eal_devargs_add(const char *devargs_str)
+rte_eal_devargs_add(const char *dev)
{
struct rte_devargs *devargs = NULL;
- const char *dev = devargs_str;
- /* use calloc instead of rte_zmalloc as it's called early at init */
devargs = calloc(1, sizeof(*devargs));
if (devargs == NULL)
goto fail;
if (rte_eal_devargs_parse(devargs, "%s", dev))
goto fail;
- TAILQ_INSERT_TAIL(&devargs_list, devargs, next);
+ if (rte_eal_devargs_insert(devargs))
+ goto fail;
return 0;
fail:
--
2.1.4
[dpdk-dev] [PATCH v2 13/18] bus: make device recognition function public
As other EAL facilities now requires the bus to be explicitly mentioned,
the function rte_bus_find_by_device_name can be used by third parties to
ease the transition to a more formal device definition scheme.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/bsdapp/eal/rte_eal_version.map | 1 +
lib/librte_eal/common/include/rte_bus.h | 12
lib/librte_eal/linuxapp/eal/rte_eal_version.map | 1 +
3 files changed, 14 insertions(+)
diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
index 0d693c8..13c8450 100644
--- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
@@ -223,6 +223,7 @@ EXPERIMENTAL {
DPDK_17.11 {
global:
+ rte_bus_find_by_device_name;
rte_bus_get_iommu_class;
rte_eal_iova_mode;
rte_eal_mbuf_default_mempool_ops;
diff --git a/lib/librte_eal/common/include/rte_bus.h
b/lib/librte_eal/common/include/rte_bus.h
index 1cae96e..6d3fb70 100644
--- a/lib/librte_eal/common/include/rte_bus.h
+++ b/lib/librte_eal/common/include/rte_bus.h
@@ -364,6 +364,18 @@ struct rte_bus *rte_bus_find_by_device(const struct
rte_device *dev);
struct rte_bus *rte_bus_find_by_name(const char *busname);
/**
+ * Find a bus capable of identifying a device.
+ *
+ * @param str
+ * A device identifier (PCI address, virtual PMD name, ...).
+ *
+ * @return
+ * A valid bus handle if found.
+ * NULL if no bus is able to parse this device.
+ */
+struct rte_bus *rte_bus_find_by_device_name(const char *str);
+
+/**
* Helper for Bus registration.
* The constructor has higher priority than PMD constructors.
*/
diff --git a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
index 9c0251e..a6fe25c 100644
--- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
@@ -227,6 +227,7 @@ EXPERIMENTAL {
DPDK_17.11 {
global:
+ rte_bus_find_by_device_name;
rte_bus_get_iommu_class;
rte_eal_iova_mode;
rte_eal_mbuf_default_mempool_ops;
--
2.1.4
[dpdk-dev] [PATCH v2 12/18] eal: add generic device declaration parameter
Add a new generic device declaration parameter:
--dev=
That allows to declare device from any bus. The format is as follows:
device_declaration := [,arg_list]
bus := bus name
c:= arbitrary character separator
device := device name (PCI location, virtual PMD name, ...)
arg_list := key value list: key1=val1[,key2=val2[,...]]
The bus name is mandatory. The character separator can be anything.
The device name is mandatory. The argument list is optional.
Examples:
--dev=pci::05:00.0,port=1
--dev=vdev_net_ring0
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/common/eal_common_options.c | 19 +++
lib/librte_eal/common/eal_options.h| 2 ++
2 files changed, 21 insertions(+)
diff --git a/lib/librte_eal/common/eal_common_options.c
b/lib/librte_eal/common/eal_common_options.c
index 603df27..b7591fd 100644
--- a/lib/librte_eal/common/eal_common_options.c
+++ b/lib/librte_eal/common/eal_common_options.c
@@ -95,6 +95,7 @@ eal_long_options[] = {
{OPT_PROC_TYPE, 1, NULL, OPT_PROC_TYPE_NUM},
{OPT_SOCKET_MEM,1, NULL, OPT_SOCKET_MEM_NUM },
{OPT_SYSLOG,1, NULL, OPT_SYSLOG_NUM },
+ {OPT_DEV, 1, NULL, OPT_DEV_NUM },
{OPT_VDEV, 1, NULL, OPT_VDEV_NUM },
{OPT_VFIO_INTR, 1, NULL, OPT_VFIO_INTR_NUM},
{OPT_VMWARE_TSC_MAP,0, NULL, OPT_VMWARE_TSC_MAP_NUM },
@@ -1120,6 +1121,21 @@ eal_parse_common_option(int opt, const char *optarg,
}
break;
+ case OPT_DEV_NUM: {
+ struct rte_devargs da;
+ int ret;
+
+ if (rte_eal_devargs_parse(&da, optarg) < 0)
+ return -1;
+ ret = rte_bus_probe_mode_set(da.bus->name,
+ RTE_BUS_PROBE_WHITELIST);
+ if (ret < 0 && ret != -ENOTSUP)
+ return -1;
+ if (eal_option_device_add(NULL, optarg) < 0)
+ return -1;
+ }
+ break;
+
case OPT_VDEV_NUM:
if (eal_option_device_add("vdev", optarg) < 0)
return -1;
@@ -1271,6 +1287,9 @@ eal_common_usage(void)
" -n CHANNELS Number of memory channels\n"
" -m MB Memory to allocate (see also
--"OPT_SOCKET_MEM")\n"
" -r RANKSForce number of memory ranks (don't
detect)\n"
+ " --"OPT_DEV" Declare a device.\n"
+ " The argument format is
[,key=val,...]\n"
+ " ex: pci:00:00.0,key=val\n"
" -b, --"OPT_PCI_BLACKLIST" Add a PCI device in black list.\n"
" Prevent EAL from using this PCI device.
The argument\n"
" format is .\n"
diff --git a/lib/librte_eal/common/eal_options.h
b/lib/librte_eal/common/eal_options.h
index 30e6bb4..d50eff7 100644
--- a/lib/librte_eal/common/eal_options.h
+++ b/lib/librte_eal/common/eal_options.h
@@ -77,6 +77,8 @@ enum {
OPT_SOCKET_MEM_NUM,
#define OPT_SYSLOG"syslog"
OPT_SYSLOG_NUM,
+#define OPT_DEV "dev"
+ OPT_DEV_NUM,
#define OPT_VDEV "vdev"
OPT_VDEV_NUM,
#define OPT_VFIO_INTR "vfio-intr"
--
2.1.4
[dpdk-dev] [PATCH v2 14/18] net/failsafe: keep legacy sub-device declaration
Allow users to use the same sub-device declaration format.
The rte_devargs library now requires the bus name to be prepended to
device declarations. While it is possible to use this new format, the
transition to the new one can be made smoother.
Signed-off-by: Gaetan Rivet
---
drivers/net/failsafe/failsafe_args.c | 11 ++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/drivers/net/failsafe/failsafe_args.c
b/drivers/net/failsafe/failsafe_args.c
index 08ce4ad..0c98264 100644
--- a/drivers/net/failsafe/failsafe_args.c
+++ b/drivers/net/failsafe/failsafe_args.c
@@ -34,6 +34,7 @@
#include
#include
+#include
#include
#include
#include
@@ -83,12 +84,20 @@ closing_paren(const char *text)
static int
fs_parse_device(struct sub_device *sdev, char *args)
{
+ struct rte_bus *bus;
struct rte_devargs *d;
int ret;
d = &sdev->devargs;
DEBUG("%s", args);
- ret = rte_eal_devargs_parse(d, "%s", args);
+ bus = rte_bus_find_by_device_name(args);
+ if (bus == NULL)
+ /* args may contain the bus name */
+ ret = rte_eal_devargs_parse(d, "%s", args);
+ else
+ /* args is a device name */
+ ret = rte_eal_devargs_parse(d, "%s:%s",
+ bus->name, args);
if (ret) {
DEBUG("devargs parsing failed with code %d", ret);
return ret;
--
2.1.4
[dpdk-dev] [PATCH v2 17/18] devargs: use proper prefix
rte_eal_devargs is redundant.
Make it concise.
Signed-off-by: Gaetan Rivet
---
app/test-pmd/cmdline.c | 2 +-
drivers/net/failsafe/failsafe_args.c| 4 ++--
examples/bond/main.c| 2 +-
lib/librte_eal/bsdapp/eal/rte_eal_version.map | 12 ++--
lib/librte_eal/common/eal_common_dev.c | 8
lib/librte_eal/common/eal_common_devargs.c | 18 +-
lib/librte_eal/common/eal_common_options.c | 4 ++--
lib/librte_eal/common/eal_common_vdev.c | 4 ++--
lib/librte_eal/common/include/rte_devargs.h | 18 +-
lib/librte_eal/linuxapp/eal/rte_eal_version.map | 12 ++--
lib/librte_ether/rte_ethdev.c | 2 +-
test/test/commands.c| 2 +-
12 files changed, 44 insertions(+), 44 deletions(-)
diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index 1204604..3b338e5 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -8322,7 +8322,7 @@ static void cmd_dump_parsed(void *parsed_result,
else if (!strcmp(res->dump, "dump_mempool"))
rte_mempool_list_dump(stdout);
else if (!strcmp(res->dump, "dump_devargs"))
- rte_eal_devargs_dump(stdout);
+ rte_devargs_dump(stdout);
else if (!strcmp(res->dump, "dump_log_types"))
rte_log_dump(stdout);
}
diff --git a/drivers/net/failsafe/failsafe_args.c
b/drivers/net/failsafe/failsafe_args.c
index 0c98264..79c90cf 100644
--- a/drivers/net/failsafe/failsafe_args.c
+++ b/drivers/net/failsafe/failsafe_args.c
@@ -93,10 +93,10 @@ fs_parse_device(struct sub_device *sdev, char *args)
bus = rte_bus_find_by_device_name(args);
if (bus == NULL)
/* args may contain the bus name */
- ret = rte_eal_devargs_parse(d, "%s", args);
+ ret = rte_devargs_parse(d, "%s", args);
else
/* args is a device name */
- ret = rte_eal_devargs_parse(d, "%s:%s",
+ ret = rte_devargs_parse(d, "%s:%s",
bus->name, args);
if (ret) {
DEBUG("devargs parsing failed with code %d", ret);
diff --git a/examples/bond/main.c b/examples/bond/main.c
index cb2..420ba1f 100644
--- a/examples/bond/main.c
+++ b/examples/bond/main.c
@@ -744,7 +744,7 @@ main(int argc, char *argv[])
/* init EAL */
ret = rte_eal_init(argc, argv);
- rte_eal_devargs_dump(stdout);
+ rte_devargs_dump(stdout);
if (ret < 0)
rte_exit(EXIT_FAILURE, "Error with EAL initialization\n");
argc -= ret;
diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
index aae8d32..0913a52 100644
--- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
@@ -19,8 +19,6 @@ DPDK_2.0 {
rte_dump_tailq;
rte_eal_alarm_cancel;
rte_eal_alarm_set;
- rte_eal_devargs_add;
- rte_eal_devargs_dump;
rte_eal_get_configuration;
rte_eal_get_lcore_state;
rte_eal_get_physmem_layout;
@@ -186,10 +184,12 @@ DPDK_17.08 {
EXPERIMENTAL {
global:
- rte_eal_devargs_insert;
- rte_eal_devargs_next;
- rte_eal_devargs_parse;
- rte_eal_devargs_remove;
+ rte_devargs_add;
+ rte_devargs_dump;
+ rte_devargs_insert;
+ rte_devargs_next;
+ rte_devargs_parse;
+ rte_devargs_remove;
rte_eal_hotplug_add;
rte_eal_hotplug_remove;
rte_service_component_register;
diff --git a/lib/librte_eal/common/eal_common_dev.c
b/lib/librte_eal/common/eal_common_dev.c
index b965e56..5608690 100644
--- a/lib/librte_eal/common/eal_common_dev.c
+++ b/lib/librte_eal/common/eal_common_dev.c
@@ -151,12 +151,12 @@ int rte_eal_hotplug_add(const char *busname, const char
*devname,
if (da == NULL)
return -ENOMEM;
- ret = rte_eal_devargs_parse(da, "%s:%s,%s",
+ ret = rte_devargs_parse(da, "%s:%s,%s",
busname, devname, devargs);
if (ret)
goto err_devarg;
- ret = rte_eal_devargs_insert(da);
+ ret = rte_devargs_insert(da);
if (ret)
goto err_devarg;
@@ -181,7 +181,7 @@ int rte_eal_hotplug_add(const char *busname, const char
*devname,
return 0;
err_devarg:
- if (rte_eal_devargs_remove(busname, devname)) {
+ if (rte_devargs_remove(busname, devname)) {
free(da->args);
free(da);
}
@@ -216,6 +216,6 @@ int rte_eal_hotplug_remove(const char *busname, const char
*devname)
if (ret)
RTE_LOG(ERR, EAL, "Driver cannot detach the device (%s)\n",
dev->name);
- rte_eal_devargs_remove(busname, devname);
+ rte_devargs_remove(busname, devna
[dpdk-dev] [PATCH v2 16/18] devargs: remove old devargs parsing function
This functionality is already covered by rte_eal_devargs_parse.
Signed-off-by: Gaetan Rivet
---
lib/librte_eal/bsdapp/eal/rte_eal_version.map | 1 -
lib/librte_eal/common/eal_common_devargs.c | 30 -
lib/librte_eal/common/include/rte_devargs.h | 28 ---
lib/librte_eal/linuxapp/eal/rte_eal_version.map | 1 -
4 files changed, 60 deletions(-)
diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
index 13c8450..aae8d32 100644
--- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
@@ -32,7 +32,6 @@ DPDK_2.0 {
rte_eal_lcore_role;
rte_eal_mp_remote_launch;
rte_eal_mp_wait_lcore;
- rte_eal_parse_devargs_str;
rte_eal_process_type;
rte_eal_remote_launch;
rte_eal_tailq_lookup;
diff --git a/lib/librte_eal/common/eal_common_devargs.c
b/lib/librte_eal/common/eal_common_devargs.c
index 1d87cd9..82c8573 100644
--- a/lib/librte_eal/common/eal_common_devargs.c
+++ b/lib/librte_eal/common/eal_common_devargs.c
@@ -53,36 +53,6 @@ TAILQ_HEAD(rte_devargs_list, rte_devargs);
struct rte_devargs_list devargs_list =
TAILQ_HEAD_INITIALIZER(devargs_list);
-int
-rte_eal_parse_devargs_str(const char *devargs_str,
- char **drvname, char **drvargs)
-{
- char *sep;
-
- if ((devargs_str) == NULL || (drvname) == NULL || (drvargs == NULL))
- return -1;
-
- *drvname = strdup(devargs_str);
- if (*drvname == NULL)
- return -1;
-
- /* set the first ',' to '\0' to split name and arguments */
- sep = strchr(*drvname, ',');
- if (sep != NULL) {
- sep[0] = '\0';
- *drvargs = strdup(sep + 1);
- } else {
- *drvargs = strdup("");
- }
-
- if (*drvargs == NULL) {
- free(*drvname);
- *drvname = NULL;
- return -1;
- }
- return 0;
-}
-
static int
bus_name_cmp(const struct rte_bus *bus, const void *name)
{
diff --git a/lib/librte_eal/common/include/rte_devargs.h
b/lib/librte_eal/common/include/rte_devargs.h
index 1fe03d6..499f7e3 100644
--- a/lib/librte_eal/common/include/rte_devargs.h
+++ b/lib/librte_eal/common/include/rte_devargs.h
@@ -74,34 +74,6 @@ struct rte_devargs {
};
/**
- * Parse a devargs string.
- *
- * For PCI devices, the format of arguments string is "PCI_ADDR" or
- * "PCI_ADDR,key=val,key2=val2,...". Examples: "08:00.1", ":5:00.0",
- * "04:00.0,arg=val".
- *
- * For virtual devices, the format of arguments string is "DRIVER_NAME*"
- * or "DRIVER_NAME*,key=val,key2=val2,...". Examples: "net_ring",
- * "net_ring0", "net_pmdAnything,arg=0:arg2=1".
- *
- * The function parses the arguments string to get driver name and driver
- * arguments.
- *
- * @param devargs_str
- * The arguments as given by the user.
- * @param drvname
- * The pointer to the string to store parsed driver name.
- * @param drvargs
- * The pointer to the string to store parsed driver arguments.
- *
- * @return
- * - 0 on success
- * - A negative value on error
- */
-int rte_eal_parse_devargs_str(const char *devargs_str,
- char **drvname, char **drvargs);
-
-/**
* Parse a device string.
*
* Verify that a bus is capable of handling the device passed
diff --git a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
index a6fe25c..323f799 100644
--- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
@@ -32,7 +32,6 @@ DPDK_2.0 {
rte_eal_lcore_role;
rte_eal_mp_remote_launch;
rte_eal_mp_wait_lcore;
- rte_eal_parse_devargs_str;
rte_eal_process_type;
rte_eal_remote_launch;
rte_eal_tailq_lookup;
--
2.1.4
[dpdk-dev] [PATCH v2 15/18] ether: use new devargs parsing function
The previous one is being deprecated.
Signed-off-by: Gaetan Rivet
---
lib/librte_ether/rte_ethdev.c | 11 +++
1 file changed, 7 insertions(+), 4 deletions(-)
diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
index bb08204..618f576 100644
--- a/lib/librte_ether/rte_ethdev.c
+++ b/lib/librte_ether/rte_ethdev.c
@@ -372,21 +372,24 @@ rte_eth_dev_is_detachable(uint16_t port_id)
/* attach the new device, then store port_id of the device */
int
-rte_eth_dev_attach(const char *devargs, uint16_t *port_id)
+rte_eth_dev_attach(const char *devstr, uint16_t *port_id)
{
int ret = -1;
int current = rte_eth_dev_count();
+ struct rte_devargs devargs;
char *name = NULL;
char *args = NULL;
- if ((devargs == NULL) || (port_id == NULL)) {
+ if ((devstr == NULL) || (port_id == NULL)) {
ret = -EINVAL;
goto err;
}
- /* parse devargs, then retrieve device name and args */
- if (rte_eal_parse_devargs_str(devargs, &name, &args))
+ /* parse device, then retrieve device name and args */
+ if (rte_eal_devargs_parse(&devargs, "%s", devstr))
goto err;
+ name = devargs.name;
+ args = devargs.args;
ret = rte_eal_dev_attach(name, args);
if (ret < 0)
--
2.1.4
Re: [dpdk-dev] [PATCH v8 1/5] net/softnic: add softnic PMD
> -Original Message- > From: Thomas Monjalon [mailto:tho...@monjalon.net] > Sent: Thursday, October 12, 2017 12:18 AM > To: Singh, Jasvinder ; Dumitrescu, Cristian > > Cc: dev@dpdk.org; Yigit, Ferruh ; Lu, Wenzhuo > > Subject: Re: [dpdk-dev] [PATCH v8 1/5] net/softnic: add softnic PMD > > 10/10/2017 12:18, Jasvinder Singh: > > +ifeq ($(CONFIG_RTE_LIBRTE_SCHED),y) > > +_LDLIBS-$(CONFIG_RTE_LIBRTE_PMD_SOFTNIC) += -lrte_pmd_softnic > > +endif > > Why linking softnic only if sched is enabled? > > Please, can you fix it for RC2? Yes, will fix this. Thanks.
[dpdk-dev] [PATCH v2 18/18] doc: remove devargs deprecation notices
These actions have been enacted. Signed-off-by: Gaetan Rivet --- doc/guides/rel_notes/deprecation.rst | 13 - 1 file changed, 13 deletions(-) diff --git a/doc/guides/rel_notes/deprecation.rst b/doc/guides/rel_notes/deprecation.rst index ef2264f..23faa19 100644 --- a/doc/guides/rel_notes/deprecation.rst +++ b/doc/guides/rel_notes/deprecation.rst @@ -16,19 +16,6 @@ Deprecation Notices - ``rte_set_log_type``, replaced by ``rte_log_set_level`` - ``rte_get_log_type``, replaced by ``rte_log_get_level`` -* eal: several API and ABI changes are planned for ``rte_devargs`` in v17.11. - The format of device command line parameters will change. The bus will need - to be explicitly stated in the device declaration. The enum ``rte_devtype`` - was used to identify a bus and will disappear. - The structure ``rte_devargs`` will change. - The ``rte_devargs_list`` will be made private. - The following functions are deprecated starting from 17.08 and will either be - modified or removed in 17.11: - - - ``rte_eal_devargs_add`` - - ``rte_eal_devargs_type_count`` - - ``rte_eal_parse_devargs_str``, replaced by ``rte_eal_devargs_parse`` - * eal: An ABI change is planned for 17.11 to make DPDK aware of IOVA address translation scheme. Reference to phys address in EAL data-structure or functions may change to -- 2.1.4
Re: [dpdk-dev] [PATCH 2/2] test/reorder: fix reorder drain test
On Tue, Sep 12, 2017 at 08:36:04PM +0530, Pavan Nikhilesh wrote:
> The reorder drain test fails due to mempool corruption caused by freeing
> packet buffer twice.
>
> Fixes: d0c9b58d7156 ("app/test: new reorder unit test")
>
> Signed-off-by: Pavan Nikhilesh
> ---
> test/test/test_reorder.c | 33 +++--
> 1 file changed, 23 insertions(+), 10 deletions(-)
>
> diff --git a/test/test/test_reorder.c b/test/test/test_reorder.c
> index 4ec22ac..51c2dcd 100644
> --- a/test/test/test_reorder.c
> +++ b/test/test/test_reorder.c
> @@ -70,13 +70,15 @@ test_reorder_create(void)
> TEST_ASSERT((b == NULL) && (rte_errno == EINVAL),
> "No error on create() with NULL name");
>
> - b = rte_reorder_create("PKT", rte_socket_id(),
> REORDER_BUFFER_SIZE_INVALID);
> + b = rte_reorder_create("PKT", rte_socket_id(),
> + REORDER_BUFFER_SIZE_INVALID);
> TEST_ASSERT((b == NULL) && (rte_errno == EINVAL),
> "No error on create() with invalid buffer size param.");
>
> b = rte_reorder_create("PKT_RO1", rte_socket_id(), REORDER_BUFFER_SIZE);
> TEST_ASSERT_EQUAL(b, test_params->b,
> - "New reorder instance created with already existing
> name");
> + "New reorder instance created with already existing"
> + " name");
>
> return 0;
> }
These changes are just cosmetic and so shouldn't really be included in
this patch. Ideally, cosmetic changes should be made only when you are
already touching the affected lines anyway due to some other change.
Also, it's bad practice to split literal strings across lines, even if
the line ends up crossing the 80-char threshold. Being able to grep
error strings is more important than keeping lines short.
This applies to many of the other changes in this patch. It makes it hard to
review for the actual functional changes.
Regards,
/Bruce
Re: [dpdk-dev] [PATCH 1/2] reorder: fix ready buffers not being nulled out
On Tue, Sep 12, 2017 at 08:36:03PM +0530, Pavan Nikhilesh wrote:
> The ready buffers should be set to NULL when drained else it might
> result in double free (mempool put) when rte_reorder_free is called.
>
> Fixes: b70b56032bff ("reorder: new library")
>
> Signed-off-by: Pavan Nikhilesh ---
> lib/librte_reorder/rte_reorder.c | 1 + 1 file changed, 1 insertion(+)
>
Rather than having an addition write for each entry going through the
reorder library, it should be possible to change free function so that
it only frees entries based on the index values.
In fact, a better solution to having reorder_free just blindly free the
mbufs would be to have reorder_free hand them back to the application,
or allow reorder_free to fail if the reorder buffer is non-empty. Making
such a change would be an ABI break, though.
/Bruce
[dpdk-dev] [PATCH v6 0/4] move vdev into drivers/bus
v6: - Don't introduce the static log type for bug, instead we use dynamic log type for vdev, suggested by Shreyansh Jain. v4 & v5: - Fix issues of compiling shared library. - Remove extra symbols in drivers/bus/vdev/rte_bus_vdev_version.map. - Address checkpatch warnings. This patch set depends on: http://dpdk.org/ml/archives/dev/2017-October/077855.html This patch set is originated from below series: http://dpdk.org/ml/archives/dev/2017-September/076821.html As per previous discussions, we tend to move all bus drivers from EAL to drivers/bus/. This patch set targets vdev bus. Jianfeng Tan (4): cryptodev: remove crypto vdev init API eal: remove dependency on vdev bus/vdev: move to vdev bus to drivers/bus bus/vdev: change log type config/common_base | 5 + doc/guides/rel_notes/deprecation.rst | 5 - drivers/bus/Makefile | 3 + drivers/bus/vdev/Makefile | 55 drivers/bus/vdev/rte_bus_vdev_version.map | 8 + drivers/bus/vdev/rte_vdev.h| 153 +++ drivers/bus/vdev/vdev.c| 356 + drivers/bus/vdev/vdev_logs.h | 45 drivers/crypto/Makefile| 1 + drivers/event/Makefile | 2 +- drivers/net/Makefile | 2 +- lib/librte_cryptodev/rte_cryptodev.c | 6 - lib/librte_cryptodev/rte_cryptodev.h | 17 -- lib/librte_cryptodev/rte_cryptodev_version.map | 1 - lib/librte_eal/bsdapp/eal/Makefile | 1 - lib/librte_eal/common/Makefile | 2 +- lib/librte_eal/common/eal_common_dev.c | 22 +- lib/librte_eal/common/eal_common_vdev.c| 342 lib/librte_eal/common/include/rte_dev.h| 24 +- lib/librte_eal/common/include/rte_vdev.h | 131 - lib/librte_eal/linuxapp/eal/Makefile | 1 - mk/rte.app.mk | 1 + 22 files changed, 637 insertions(+), 546 deletions(-) create mode 100644 drivers/bus/vdev/Makefile create mode 100644 drivers/bus/vdev/rte_bus_vdev_version.map create mode 100644 drivers/bus/vdev/rte_vdev.h create mode 100644 drivers/bus/vdev/vdev.c create mode 100644 drivers/bus/vdev/vdev_logs.h delete mode 100644 lib/librte_eal/common/eal_common_vdev.c delete mode 100644 lib/librte_eal/common/include/rte_vdev.h -- 2.7.4
[dpdk-dev] [PATCH v6 1/4] cryptodev: remove crypto vdev init API
Remove rte_cryptodev_create_vdev() for duplication.
Signed-off-by: Jianfeng Tan
Acked-by: Pablo de Lara
---
doc/guides/rel_notes/deprecation.rst | 5 -
lib/librte_cryptodev/rte_cryptodev.c | 6 --
lib/librte_cryptodev/rte_cryptodev.h | 17 -
lib/librte_cryptodev/rte_cryptodev_version.map | 1 -
4 files changed, 29 deletions(-)
diff --git a/doc/guides/rel_notes/deprecation.rst
b/doc/guides/rel_notes/deprecation.rst
index f4269f0..85a9287 100644
--- a/doc/guides/rel_notes/deprecation.rst
+++ b/doc/guides/rel_notes/deprecation.rst
@@ -78,11 +78,6 @@ Deprecation Notices
``rte_cryptodev`` respectively to support security protocol offloaded
operations.
-* cryptodev: the following function is deprecated starting from 17.08 and will
- be removed in 17.11:
-
- - ``rte_cryptodev_create_vdev``
-
* cryptodev: the following function will be static in 17.11 and included
by all crypto drivers, therefore, will not be public:
diff --git a/lib/librte_cryptodev/rte_cryptodev.c
b/lib/librte_cryptodev/rte_cryptodev.c
index 5e8f7f4..f2c55cc 100644
--- a/lib/librte_cryptodev/rte_cryptodev.c
+++ b/lib/librte_cryptodev/rte_cryptodev.c
@@ -377,12 +377,6 @@ rte_cryptodev_get_feature_name(uint64_t flag)
}
}
-int
-rte_cryptodev_create_vdev(const char *name, const char *args)
-{
- return rte_vdev_init(name, args);
-}
-
struct rte_cryptodev *
rte_cryptodev_pmd_get_dev(uint8_t dev_id)
{
diff --git a/lib/librte_cryptodev/rte_cryptodev.h
b/lib/librte_cryptodev/rte_cryptodev.h
index ad97ff9..c5d6939 100644
--- a/lib/librte_cryptodev/rte_cryptodev.h
+++ b/lib/librte_cryptodev/rte_cryptodev.h
@@ -433,23 +433,6 @@ struct rte_cryptodev_stats {
/**< Max length of name of crypto PMD */
/**
- * @deprecated
- *
- * Create a virtual crypto device
- *
- * @param nameCryptodev PMD name of device to be created.
- * @param argsOptions arguments for device.
- *
- * @return
- * - On successful creation of the cryptodev the device index is returned,
- * which will be between 0 and rte_cryptodev_count().
- * - In the case of a failure, returns -1.
- */
-__rte_deprecated
-extern int
-rte_cryptodev_create_vdev(const char *name, const char *args);
-
-/**
* Get the device identifier for the named crypto device.
*
* @param namedevice name to select the device structure.
diff --git a/lib/librte_cryptodev/rte_cryptodev_version.map
b/lib/librte_cryptodev/rte_cryptodev_version.map
index abdbbda..b07f4ca 100644
--- a/lib/librte_cryptodev/rte_cryptodev_version.map
+++ b/lib/librte_cryptodev/rte_cryptodev_version.map
@@ -7,7 +7,6 @@ DPDK_16.04 {
rte_cryptodev_close;
rte_cryptodev_count;
rte_cryptodev_configure;
- rte_cryptodev_create_vdev;
rte_cryptodev_get_dev_id;
rte_cryptodev_get_feature_name;
rte_cryptodev_info_get;
--
2.7.4
[dpdk-dev] [PATCH v6 3/4] bus/vdev: move to vdev bus to drivers/bus
Move the vdev bus from lib/librte_eal to drivers/bus.
As the crypto vdev helper function refers to data structure
in rte_vdev.h, so we move those helper function into drivers/bus
too.
Signed-off-by: Jianfeng Tan
---
config/common_base| 5 +
drivers/bus/Makefile | 3 +
drivers/bus/vdev/Makefile | 55 +
drivers/bus/vdev/rte_bus_vdev_version.map | 8 +
drivers/bus/vdev/rte_vdev.h | 153 +
drivers/bus/vdev/vdev.c | 344 ++
drivers/crypto/Makefile | 1 +
drivers/event/Makefile| 2 +-
drivers/net/Makefile | 2 +-
lib/librte_eal/bsdapp/eal/Makefile| 1 -
lib/librte_eal/common/Makefile| 2 +-
lib/librte_eal/common/eal_common_vdev.c | 342 -
lib/librte_eal/common/include/rte_dev.h | 24 +--
lib/librte_eal/common/include/rte_vdev.h | 131
lib/librte_eal/linuxapp/eal/Makefile | 1 -
mk/rte.app.mk | 1 +
16 files changed, 574 insertions(+), 501 deletions(-)
create mode 100644 drivers/bus/vdev/Makefile
create mode 100644 drivers/bus/vdev/rte_bus_vdev_version.map
create mode 100644 drivers/bus/vdev/rte_vdev.h
create mode 100644 drivers/bus/vdev/vdev.c
delete mode 100644 lib/librte_eal/common/eal_common_vdev.c
delete mode 100644 lib/librte_eal/common/include/rte_vdev.h
diff --git a/config/common_base b/config/common_base
index 22068c5..c6f2aae 100644
--- a/config/common_base
+++ b/config/common_base
@@ -754,3 +754,8 @@ CONFIG_RTE_APP_CRYPTO_PERF=y
# Compile the eventdev application
#
CONFIG_RTE_APP_EVENTDEV=y
+
+#
+# Compile the vdev bus
+#
+CONFIG_RTE_LIBRTE_VDEV_BUS=y
diff --git a/drivers/bus/Makefile b/drivers/bus/Makefile
index 6cb6466..91ffa87 100644
--- a/drivers/bus/Makefile
+++ b/drivers/bus/Makefile
@@ -38,4 +38,7 @@ DEPDIRS-dpaa = $(core-libs)
DIRS-$(CONFIG_RTE_LIBRTE_FSLMC_BUS) += fslmc
DEPDIRS-fslmc = $(core-libs)
+DIRS-$(CONFIG_RTE_LIBRTE_VDEV_BUS) += vdev
+DEPDIRS-vdev = $(core-libs)
+
include $(RTE_SDK)/mk/rte.subdir.mk
diff --git a/drivers/bus/vdev/Makefile b/drivers/bus/vdev/Makefile
new file mode 100644
index 000..6c48ba0
--- /dev/null
+++ b/drivers/bus/vdev/Makefile
@@ -0,0 +1,55 @@
+# BSD LICENSE
+#
+# Copyright(c) 2017 Intel Corporation. All rights reserved.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#
+# * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+# * Redistributions in binary form must reproduce the above copyright
+# notice, this list of conditions and the following disclaimer in
+# the documentation and/or other materials provided with the
+# distribution.
+# * Neither the name of Intel Corporation nor the names of its
+# contributors may be used to endorse or promote products derived
+# from this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+include $(RTE_SDK)/mk/rte.vars.mk
+
+#
+# library name
+#
+LIB = librte_bus_vdev.a
+
+CFLAGS += -O3
+CFLAGS += $(WERROR_FLAGS)
+
+# versioning export map
+EXPORT_MAP := rte_bus_vdev_version.map
+
+# library version
+LIBABIVER := 1
+
+SRCS-y += vdev.c
+
+#
+# Export include files
+#
+SYMLINK-y-include += rte_vdev.h
+
+include $(RTE_SDK)/mk/rte.lib.mk
diff --git a/drivers/bus/vdev/rte_bus_vdev_version.map
b/drivers/bus/vdev/rte_bus_vdev_version.map
new file mode 100644
index 000..6ccb789
--- /dev/null
+++ b/drivers/bus/vdev/rte_bus_vdev_version.map
@@ -0,0 +1,8 @@
+DPDK_17.11 {
+ global:
+
+ rte_vdev_init;
+ rte_vdev_register;
+ rte_vdev_uninit;
+ rte_vdev_unregister;
+};
diff --git a/drivers/bus/vdev/rte_vdev.h b/drivers/bus/vdev/rte_vdev.h
new file mode 100644
index 000..41762b8
--- /dev/null
+++ b/drivers/bus/vdev/rte_vdev.h
@@ -0,0 +1,153 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2016 RehiveTe
[dpdk-dev] [PATCH v6 4/4] bus/vdev: change log type
Use specialized dynamic log type for vdev bus logging.
Suggested-by: Gaetan Rivet
Suggested-by: Shreyansh Jain
Signed-off-by: Jianfeng Tan
---
drivers/bus/vdev/vdev.c | 20
drivers/bus/vdev/vdev_logs.h | 45
2 files changed, 61 insertions(+), 4 deletions(-)
create mode 100644 drivers/bus/vdev/vdev_logs.h
diff --git a/drivers/bus/vdev/vdev.c b/drivers/bus/vdev/vdev.c
index a16eb71..be2be6f 100644
--- a/drivers/bus/vdev/vdev.c
+++ b/drivers/bus/vdev/vdev.c
@@ -47,6 +47,9 @@
#include
#include "rte_vdev.h"
+#include "vdev_logs.h"
+
+int vdev_logtype_bus;
/* Forward declare to access virtual bus name */
static struct rte_bus rte_vdev_bus;
@@ -103,7 +106,7 @@ vdev_probe_all_drivers(struct rte_vdev_device *dev)
name = rte_vdev_device_name(dev);
- RTE_LOG(DEBUG, EAL, "Search driver %s to probe device %s\n", name,
+ VDEV_LOG(DEBUG, "Search driver %s to probe device %s\n", name,
rte_vdev_device_name(dev));
if (vdev_parse(name, &driver))
@@ -190,7 +193,7 @@ rte_vdev_init(const char *name, const char *args)
ret = vdev_probe_all_drivers(dev);
if (ret) {
if (ret > 0)
- RTE_LOG(ERR, EAL, "no driver found for %s\n", name);
+ VDEV_LOG(ERR, "no driver found for %s\n", name);
goto fail;
}
@@ -213,7 +216,7 @@ vdev_remove_driver(struct rte_vdev_device *dev)
const struct rte_vdev_driver *driver;
if (!dev->device.driver) {
- RTE_LOG(DEBUG, EAL, "no driver attach to device %s\n", name);
+ VDEV_LOG(DEBUG, "no driver attach to device %s\n", name);
return 1;
}
@@ -294,7 +297,7 @@ vdev_probe(void)
continue;
if (vdev_probe_all_drivers(dev)) {
- RTE_LOG(ERR, EAL, "failed to initialize %s device\n",
+ VDEV_LOG(ERR, "failed to initialize %s device\n",
rte_vdev_device_name(dev));
return -1;
}
@@ -342,3 +345,12 @@ static struct rte_bus rte_vdev_bus = {
};
RTE_REGISTER_BUS(vdev, rte_vdev_bus);
+
+RTE_INIT(vdev_init_log);
+static void
+vdev_init_log(void)
+{
+ vdev_logtype_bus = rte_log_register("bus.vdev");
+ if (vdev_logtype_bus >= 0)
+ rte_log_set_level(vdev_logtype_bus, RTE_LOG_NOTICE);
+}
diff --git a/drivers/bus/vdev/vdev_logs.h b/drivers/bus/vdev/vdev_logs.h
new file mode 100644
index 000..43b0ab8
--- /dev/null
+++ b/drivers/bus/vdev/vdev_logs.h
@@ -0,0 +1,45 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2017 Intel Corporation. All rights reserved.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ * * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ * * Redistributions in binary form must reproduce the above copyright
+ * notice, this list of conditions and the following disclaimer in
+ * the documentation and/or other materials provided with the
+ * distribution.
+ * * Neither the name of Intel Corporation nor the names of its
+ * contributors may be used to endorse or promote products derived
+ * from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef _VDEV_LOGS_H_
+#define _VDEV_LOGS_H_
+
+#include
+
+extern int vdev_logtype_bus;
+
+#define VDEV_LOG(level, fmt, args...) \
+ rte_log(RTE_LOG_ ## level, vdev_logtype_bus, "%s(): " fmt "\n", \
+ __func__, ##args)
+
+#endif /* _VDEV_LOGS_H_ */
--
2.7.4
[dpdk-dev] [PATCH v6 2/4] eal: remove dependency on vdev
We can call bus->plug() to avoid calling rte_vdev_init() explicitly.
Signed-off-by: Jianfeng Tan
---
lib/librte_eal/common/eal_common_dev.c | 22 ++
1 file changed, 6 insertions(+), 16 deletions(-)
diff --git a/lib/librte_eal/common/eal_common_dev.c
b/lib/librte_eal/common/eal_common_dev.c
index e251275..dda8f58 100644
--- a/lib/librte_eal/common/eal_common_dev.c
+++ b/lib/librte_eal/common/eal_common_dev.c
@@ -67,7 +67,6 @@ static int cmp_dev_name(const struct rte_device *dev, const
void *_name)
int rte_eal_dev_attach(const char *name, const char *devargs)
{
struct rte_bus *bus;
- int ret;
if (name == NULL || devargs == NULL) {
RTE_LOG(ERR, EAL, "Invalid device or arguments provided\n");
@@ -80,22 +79,13 @@ int rte_eal_dev_attach(const char *name, const char
*devargs)
name);
return -EINVAL;
}
- if (strcmp(bus->name, "pci") == 0)
- return rte_eal_hotplug_add("pci", name, devargs);
- if (strcmp(bus->name, "vdev") != 0) {
- RTE_LOG(ERR, EAL, "Device attach is only supported for PCI and
vdev devices.\n");
- return -ENOTSUP;
- }
+ if (strcmp(bus->name, "pci") == 0 || strcmp(bus->name, "vdev") == 0)
+ return rte_eal_hotplug_add(bus->name, name, devargs);
- /*
-* If we haven't found a bus device the user meant to "hotplug" a
-* virtual device instead.
-*/
- ret = rte_vdev_init(name, devargs);
- if (ret)
- RTE_LOG(ERR, EAL, "Driver cannot attach the device (%s)\n",
- name);
- return ret;
+ RTE_LOG(ERR, EAL,
+ "Device attach is only supported for PCI and vdev devices.\n");
+
+ return -ENOTSUP;
}
int rte_eal_dev_detach(struct rte_device *dev)
--
2.7.4
Re: [dpdk-dev] [PATCH v2 5/5] eal/timer: honor architecture specific rdtsc hz function
On Wed, Oct 11, 2017 at 09:25:58PM +0200, Thomas Monjalon wrote:
> 11/10/2017 20:57, Jerin Jacob:
> > From: Thomas Monjalon
> > > 22/09/2017 10:25, Gowrishankar:
> > > > From: Jerin Jacob
> > > >
> > > > When calibrating the tsc frequency, first, probe the architecture
> > > > specific
> > > > rdtsc hz function. if not available, use the existing calibrate scheme
> > > > to calibrate the tsc frequency.
> > > >
> > > > Signed-off-by: Jerin Jacob
> > >
> > > I agree on the idea.
> >
> > OK
> >
> > > The namespace of cycles related function in DPDK is a real mess.
> >
> > Absolutely!!
> >
> > > I think we can choose better names in this series as a first step
> > > to tidy this mess.
> > > I will explain below.
> > >
> > > At first, we should avoid TSC and RDTSC which are Intel-only wording.
> > > The generic word could be "cycles" (the word used in arch headers),
> > > or "ticks".
> > > We should also name the timer sources or their function in a generic way.
> > > Examples: CPU cycles? fast counter? precise counter?
> > >
> > > Sometimes we use "hz", sometimes "freq".
> > > It would better to keep one of them.
> > >
> > > > --- a/lib/librte_eal/common/eal_common_timer.c
> > > > +++ b/lib/librte_eal/common/eal_common_timer.c
> > > > @@ -80,8 +80,11 @@
> > > > void
> > > > set_tsc_freq(void)
> > > > {
> > > > - uint64_t freq = get_tsc_freq();
> > > > + uint64_t freq;
> > > >
> > > > + freq = rte_rdtsc_arch_hz();
> > >
> > > This new function is arch-specific and exported as a new API.
> >
> > I thought of avoid exporting it. But then if the function is in
> > lib/librte_eal/common/include/arch/../rte_cycles.h it is anyway exposed to
> > application. i.e whatever files in lib/librte_eal/common/include/arch/../
> > anyway exposed to application.
>
> Ah yes, you are right!
>
> > See last comment.
> >
> > >
> > > > + if (!freq)
> > > > + freq = get_tsc_freq();
> > >
> > > The function get_tsc_freq is guessing the freq with OS-specific method.
> > >
> > > > if (!freq)
> > > > freq = estimate_tsc_freq();
> > >
> > > The function estimate_tsc_freq is doing an estimation based on sleep().
> > >
> > > At the end, the most accurate frequency is saved in eal_tsc_resolution_hz
> > > and can be retrieved with rte_get_tsc_hz().
> > > I don't understand why rte_rdtsc_arch_hz() is also exported to the apps.
> > >
> > > TSC and HPET timer sources are wrapped in rte_get_timer_hz() in the
> > > Similarly we can get the current timer with rte_get_timer_cycles().
> > > In the case of TSC, it calls rte_get_tsc_cycles() which is an alias
> > > of rte_rdtsc().
> > > Some code is still using directly rte_rdtsc().
> > > There is also rte_rdtsc_precise which adds a memory barrier.
> > >
> > > The real question is what is the right abstraction for the application?
> > > Do we want the fastest timer? the CPU timer? a precise timer?
> > >
> > > I would like to see a real discussion on this topic, in order of building
> > > a new timer API which would alias the old one for some time.
> >
> > I guess, we may need to see to how abstract vmware TSC support also in
> > proper way
>
> Yes
>
> > > If you don't want to bother with all these questions, I suggest to not
> > > export the new function rte_rdtsc_arch_hz() and rename it to tsc_arch_hz.
> >
> > If I understand it correctly, You would like to create a header file
> > in lib/librte_eal/common/include/arch/../ which should not be exported and
> > change
> > the name to tsc_arch_hz.
>
> I had not think about the way to do this.
> What about having internal headers in lib/librte_eal/common/arch/ ?
>
Yes, this area needs cleanup, but I also think that this patchset (and
follow-on patch for x86-specific implementation) does not make things
significantly worse than they are now, while also giving significant
benefits for users both in terms of improved clock accuracy and reduced
startup time (due to not needing sleep). Therefore I'd like to see this
merged into 17.11 as a definite improvement, even if it does not "fix" the
bigger issues of poor naming etc. I think this is important enough an
improvement that we need to see it in the LTS release.
/Bruce
[dpdk-dev] [PATCH] remove unused MRVL_DEBUG config option
Remove unused MRVL_DEBUG configuration option and update driver documentation accordingly. Signed-off-by: Tomasz Duszynski --- config/common_base | 1 - doc/guides/nics/mrvl.rst | 4 2 files changed, 5 deletions(-) diff --git a/config/common_base b/config/common_base index dc4da9a..ddfd191 100644 --- a/config/common_base +++ b/config/common_base @@ -264,7 +264,6 @@ CONFIG_RTE_LIBRTE_NFP_DEBUG=n # Compile Marvell PMD driver # CONFIG_RTE_LIBRTE_MRVL_PMD=n -CONFIG_RTE_LIBRTE_MRVL_DEBUG=n CONFIG_RTE_MRVL_MUSDK_DMA_MEMSIZE=41943040 # diff --git a/doc/guides/nics/mrvl.rst b/doc/guides/nics/mrvl.rst index 462bc0e..8fde273 100644 --- a/doc/guides/nics/mrvl.rst +++ b/doc/guides/nics/mrvl.rst @@ -122,10 +122,6 @@ The following options can be modified in the ``config`` file. Toggle compilation of the librte_pmd_mrvl driver. -- ``CONFIG_RTE_LIBRTE_MRVL_DEBUG`` (default ``n``) - -Toggle display of debugging messages. - - ``CONFIG_RTE_MRVL_MUSDK_DMA_MEMSIZE`` (default ``41943040``) Size in bytes of the contiguous memory region that MUSDK will allocate -- 2.7.4
[dpdk-dev] [PATCH 0/6] fix port id type
Now ethdev port_id has been increased to uint16_t from uint8_t in the below patch http://www.dpdk.org/dev/patchwork/patch/29418/ but 1. Some features which are applied were still developed based on older version uint8_t port_id. 2. Some places using port id such as testpmd are omitted in previous patch. The patchset fixes some port id type issues. Zhiyong Yang (6): lib/librte_ether: fix port id type net/bnxt: fix port id type net/liquidio: fix port id type app/pdump: fix port id type app/proc_info: fix port id type app/testpmd: fix port id type app/pdump/main.c | 4 +- app/proc_info/main.c | 14 +- app/test-pmd/cmdline.c| 427 +++--- app/test-pmd/config.c | 14 +- app/test-pmd/testpmd.c| 13 +- app/test-pmd/testpmd.h| 16 +- drivers/net/bnxt/bnxt_hwrm.c | 2 +- drivers/net/liquidio/lio_struct.h | 2 +- lib/librte_ether/ethdev_profile.c | 6 +- lib/librte_ether/ethdev_profile.h | 2 +- lib/librte_ether/rte_ethdev.c | 2 +- lib/librte_ether/rte_ethdev.h | 2 +- 12 files changed, 253 insertions(+), 251 deletions(-) -- 2.13.3
[dpdk-dev] [PATCH 1/6] lib/librte_ether: fix port id type
Some features applied were still developed based on older version uint8_t
port_id, but port_id has been increased range to uint16_t. The patch fixes
the issue.
Fixes: f8244c6399d9 ("ethdev: increase port id range")
Signed-off-by: Zhiyong Yang
---
lib/librte_ether/ethdev_profile.c | 6 +++---
lib/librte_ether/ethdev_profile.h | 2 +-
lib/librte_ether/rte_ethdev.c | 2 +-
lib/librte_ether/rte_ethdev.h | 2 +-
4 files changed, 6 insertions(+), 6 deletions(-)
diff --git a/lib/librte_ether/ethdev_profile.c
b/lib/librte_ether/ethdev_profile.c
index 13c48a8f3..8792ed56e 100644
--- a/lib/librte_ether/ethdev_profile.c
+++ b/lib/librte_ether/ethdev_profile.c
@@ -82,7 +82,7 @@ struct itt_profile_rx_data itt_rx_data[RTE_MAX_ETHPORTS];
* type declaration.
*/
static uint16_t
-collect_itt_rx_burst_cb(uint8_t port_id, uint16_t queue_id,
+collect_itt_rx_burst_cb(uint16_t port_id, uint16_t queue_id,
__rte_unused struct rte_mbuf *pkts[], uint16_t nb_pkts,
__rte_unused uint16_t max_pkts, __rte_unused void *user_param)
{
@@ -121,7 +121,7 @@ collect_itt_rx_burst_cb(uint8_t port_id, uint16_t queue_id,
* - On failure, a negative value.
*/
static inline int
-itt_profile_rx_init(uint8_t port_id, char *port_name, uint8_t rx_queue_num)
+itt_profile_rx_init(uint16_t port_id, char *port_name, uint8_t rx_queue_num)
{
uint16_t q_id;
@@ -155,7 +155,7 @@ itt_profile_rx_init(uint8_t port_id, char *port_name,
uint8_t rx_queue_num)
#endif /* RTE_ETHDEV_PROFILE_ITT_WASTED_RX_ITERATIONS */
int
-__rte_eth_profile_rx_init(__rte_unused uint8_t port_id,
+__rte_eth_profile_rx_init(__rte_unused uint16_t port_id,
__rte_unused struct rte_eth_dev *dev)
{
#ifdef RTE_ETHDEV_PROFILE_ITT_WASTED_RX_ITERATIONS
diff --git a/lib/librte_ether/ethdev_profile.h
b/lib/librte_ether/ethdev_profile.h
index 32f103ec1..697facff7 100644
--- a/lib/librte_ether/ethdev_profile.h
+++ b/lib/librte_ether/ethdev_profile.h
@@ -51,6 +51,6 @@
* - On failure, a negative value.
*/
int
-__rte_eth_profile_rx_init(uint8_t port_id, struct rte_eth_dev *dev);
+__rte_eth_profile_rx_init(uint16_t port_id, struct rte_eth_dev *dev);
#endif
diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c
index 9460161eb..8f027c71b 100644
--- a/lib/librte_ether/rte_ethdev.c
+++ b/lib/librte_ether/rte_ethdev.c
@@ -3628,7 +3628,7 @@ rte_eth_dev_adjust_nb_rx_tx_desc(uint16_t port_id,
}
int
-rte_eth_dev_pool_ops_supported(uint8_t port_id, const char *pool)
+rte_eth_dev_pool_ops_supported(uint16_t port_id, const char *pool)
{
struct rte_eth_dev *dev;
diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h
index 354d170ef..3f19117b2 100644
--- a/lib/librte_ether/rte_ethdev.h
+++ b/lib/librte_ether/rte_ethdev.h
@@ -4583,7 +4583,7 @@ int rte_eth_dev_adjust_nb_rx_tx_desc(uint16_t port_id,
* - -EINVAL: Pool param is null.
*/
int
-rte_eth_dev_pool_ops_supported(uint8_t port_id, const char *pool);
+rte_eth_dev_pool_ops_supported(uint16_t port_id, const char *pool);
#ifdef __cplusplus
}
--
2.13.3
[dpdk-dev] [PATCH 2/6] net/bnxt: fix port id type
Fixes: f8244c6399d9 ("ethdev: increase port id range")
Signed-off-by: Zhiyong Yang
---
drivers/net/bnxt/bnxt_hwrm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/bnxt/bnxt_hwrm.c b/drivers/net/bnxt/bnxt_hwrm.c
index c9b04e248..0e96d3c4d 100644
--- a/drivers/net/bnxt/bnxt_hwrm.c
+++ b/drivers/net/bnxt/bnxt_hwrm.c
@@ -1940,7 +1940,7 @@ static uint16_t bnxt_parse_eth_link_speed(uint32_t
conf_link_speed)
ETH_LINK_SPEED_10G | ETH_LINK_SPEED_20G | ETH_LINK_SPEED_25G | \
ETH_LINK_SPEED_40G | ETH_LINK_SPEED_50G)
-static int bnxt_valid_link_speed(uint32_t link_speed, uint8_t port_id)
+static int bnxt_valid_link_speed(uint32_t link_speed, uint16_t port_id)
{
uint32_t one_speed;
--
2.13.3
[dpdk-dev] [PATCH 3/6] net/liquidio: fix port id type
port_id in struct lio_device should be increased range to uint16_t since
port id in rte_eth_dev_data has already been defined as uint16_t.
Fixes: f8244c6399d9 ("ethdev: increase port id range")
Signed-off-by: Zhiyong Yang
---
drivers/net/liquidio/lio_struct.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/liquidio/lio_struct.h
b/drivers/net/liquidio/lio_struct.h
index 635e47fbf..10e3976af 100644
--- a/drivers/net/liquidio/lio_struct.h
+++ b/drivers/net/liquidio/lio_struct.h
@@ -684,7 +684,7 @@ struct lio_device {
uint8_t nb_tx_queues;
uint8_t port_configured;
struct lio_rss_ctx rss_state;
- uint8_t port_id;
+ uint16_t port_id;
char firmware_version[LIO_FW_VERSION_LENGTH];
};
#endif /* _LIO_STRUCT_H_ */
--
2.13.3
[dpdk-dev] [PATCH 5/6] app/proc_info: fix port id type
Fixes: f8244c6399d9 ("ethdev: increase port id range")
Signed-off-by: Zhiyong Yang
---
app/proc_info/main.c | 14 +++---
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/app/proc_info/main.c b/app/proc_info/main.c
index 12566e27d..64fbbd0f8 100644
--- a/app/proc_info/main.c
+++ b/app/proc_info/main.c
@@ -310,7 +310,7 @@ meminfo_display(void)
}
static void
-nic_stats_display(uint8_t port_id)
+nic_stats_display(uint16_t port_id)
{
struct rte_eth_stats stats;
uint8_t i;
@@ -349,7 +349,7 @@ nic_stats_display(uint8_t port_id)
}
static void
-nic_stats_clear(uint8_t port_id)
+nic_stats_clear(uint16_t port_id)
{
printf("\n Clearing NIC stats for port %d\n", port_id);
rte_eth_stats_reset(port_id);
@@ -411,7 +411,7 @@ static void collectd_resolve_cnt_type(char *cnt_type,
size_t cnt_type_len,
}
static void
-nic_xstats_by_name_display(uint8_t port_id, char *name)
+nic_xstats_by_name_display(uint16_t port_id, char *name)
{
uint64_t id;
@@ -426,7 +426,7 @@ nic_xstats_by_name_display(uint8_t port_id, char *name)
}
static void
-nic_xstats_by_ids_display(uint8_t port_id, uint64_t *ids, int len)
+nic_xstats_by_ids_display(uint16_t port_id, uint64_t *ids, int len)
{
struct rte_eth_xstat_name *xstats_names;
uint64_t *values;
@@ -473,7 +473,7 @@ nic_xstats_by_ids_display(uint8_t port_id, uint64_t *ids,
int len)
}
static void
-nic_xstats_display(uint8_t port_id)
+nic_xstats_display(uint16_t port_id)
{
struct rte_eth_xstat_name *xstats_names;
uint64_t *values;
@@ -541,7 +541,7 @@ nic_xstats_display(uint8_t port_id)
}
static void
-nic_xstats_clear(uint8_t port_id)
+nic_xstats_clear(uint16_t port_id)
{
printf("\n Clearing NIC xstats for port %d\n", port_id);
rte_eth_xstats_reset(port_id);
@@ -618,7 +618,7 @@ main(int argc, char **argv)
char n_flag[] = "-n4";
char mp_flag[] = "--proc-type=secondary";
char *argp[argc + 3];
- uint8_t nb_ports;
+ uint16_t nb_ports;
/* preparse app arguments */
ret = proc_info_preparse_args(argc, argv);
--
2.13.3
[dpdk-dev] [PATCH 4/6] app/pdump: fix port id type
Increase port id range to 16 bits and remove the unnecessary cast.
Fixes: f8244c6399d9 ("ethdev: increase port id range")
Signed-off-by: Zhiyong Yang
---
app/pdump/main.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/app/pdump/main.c b/app/pdump/main.c
index 090a50cfc..3d8e50273 100644
--- a/app/pdump/main.c
+++ b/app/pdump/main.c
@@ -579,7 +579,7 @@ signal_handler(int sig_num)
}
static inline int
-configure_vdev(uint8_t port_id)
+configure_vdev(uint16_t port_id)
{
struct ether_addr addr;
const uint16_t rxRings = 0, txRings = 1;
@@ -609,7 +609,7 @@ configure_vdev(uint8_t port_id)
rte_eth_macaddr_get(port_id, &addr);
printf("Port %u MAC: %02"PRIx8" %02"PRIx8" %02"PRIx8
" %02"PRIx8" %02"PRIx8" %02"PRIx8"\n",
- (unsigned)port_id,
+ port_id,
addr.addr_bytes[0], addr.addr_bytes[1],
addr.addr_bytes[2], addr.addr_bytes[3],
addr.addr_bytes[4], addr.addr_bytes[5]);
--
2.13.3
[dpdk-dev] [PATCH 6/6] app/testpmd: fix port id type
Fixes: f8244c6399d9 ("ethdev: increase port id range")
Signed-off-by: Zhiyong Yang
---
app/test-pmd/cmdline.c | 427 +
app/test-pmd/config.c | 14 +-
app/test-pmd/testpmd.c | 13 +-
app/test-pmd/testpmd.h | 16 +-
4 files changed, 236 insertions(+), 234 deletions(-)
diff --git a/app/test-pmd/cmdline.c b/app/test-pmd/cmdline.c
index 120460452..bb19d726f 100644
--- a/app/test-pmd/cmdline.c
+++ b/app/test-pmd/cmdline.c
@@ -1141,7 +1141,7 @@ cmdline_parse_inst_t cmd_operate_attach_port = {
struct cmd_operate_detach_port_result {
cmdline_fixed_string_t port;
cmdline_fixed_string_t keyword;
- uint8_t port_id;
+ portid_t port_id;
};
static void cmd_operate_detach_port_parsed(void *parsed_result,
@@ -1164,7 +1164,7 @@ cmdline_parse_token_string_t
cmd_operate_detach_port_keyword =
keyword, "detach");
cmdline_parse_token_num_t cmd_operate_detach_port_port_id =
TOKEN_NUM_INITIALIZER(struct cmd_operate_detach_port_result,
- port_id, UINT8);
+ port_id, UINT16);
cmdline_parse_inst_t cmd_operate_detach_port = {
.f = cmd_operate_detach_port_parsed,
@@ -1543,7 +1543,7 @@ struct cmd_config_mtu_result {
cmdline_fixed_string_t port;
cmdline_fixed_string_t keyword;
cmdline_fixed_string_t mtu;
- uint8_t port_id;
+ portid_t port_id;
uint16_t value;
};
@@ -1571,7 +1571,7 @@ cmdline_parse_token_string_t cmd_config_mtu_mtu =
TOKEN_STRING_INITIALIZER(struct cmd_config_mtu_result, keyword,
"mtu");
cmdline_parse_token_num_t cmd_config_mtu_port_id =
- TOKEN_NUM_INITIALIZER(struct cmd_config_mtu_result, port_id, UINT8);
+ TOKEN_NUM_INITIALIZER(struct cmd_config_mtu_result, port_id, UINT16);
cmdline_parse_token_num_t cmd_config_mtu_value =
TOKEN_NUM_INITIALIZER(struct cmd_config_mtu_result, value, UINT16);
@@ -1825,7 +1825,7 @@ cmdline_parse_inst_t cmd_config_rss = {
struct cmd_config_rss_hash_key {
cmdline_fixed_string_t port;
cmdline_fixed_string_t config;
- uint8_t port_id;
+ portid_t port_id;
cmdline_fixed_string_t rss_hash_key;
cmdline_fixed_string_t rss_type;
cmdline_fixed_string_t key;
@@ -1907,7 +1907,7 @@ cmdline_parse_token_string_t
cmd_config_rss_hash_key_config =
TOKEN_STRING_INITIALIZER(struct cmd_config_rss_hash_key, config,
"config");
cmdline_parse_token_num_t cmd_config_rss_hash_key_port_id =
- TOKEN_NUM_INITIALIZER(struct cmd_config_rss_hash_key, port_id, UINT8);
+ TOKEN_NUM_INITIALIZER(struct cmd_config_rss_hash_key, port_id, UINT16);
cmdline_parse_token_string_t cmd_config_rss_hash_key_rss_hash_key =
TOKEN_STRING_INITIALIZER(struct cmd_config_rss_hash_key,
rss_hash_key, "rss-hash-key");
@@ -2037,7 +2037,7 @@ cmdline_parse_inst_t cmd_config_rxtx_queue = {
struct cmd_config_rss_reta {
cmdline_fixed_string_t port;
cmdline_fixed_string_t keyword;
- uint8_t port_id;
+ portid_t port_id;
cmdline_fixed_string_t name;
cmdline_fixed_string_t list_name;
cmdline_fixed_string_t list_of_items;
@@ -2146,7 +2146,7 @@ cmdline_parse_token_string_t cmd_config_rss_reta_port =
cmdline_parse_token_string_t cmd_config_rss_reta_keyword =
TOKEN_STRING_INITIALIZER(struct cmd_config_rss_reta, keyword, "config");
cmdline_parse_token_num_t cmd_config_rss_reta_port_id =
- TOKEN_NUM_INITIALIZER(struct cmd_config_rss_reta, port_id, UINT8);
+ TOKEN_NUM_INITIALIZER(struct cmd_config_rss_reta, port_id, UINT16);
cmdline_parse_token_string_t cmd_config_rss_reta_name =
TOKEN_STRING_INITIALIZER(struct cmd_config_rss_reta, name, "rss");
cmdline_parse_token_string_t cmd_config_rss_reta_list_name =
@@ -2173,7 +2173,7 @@ cmdline_parse_inst_t cmd_config_rss_reta = {
struct cmd_showport_reta {
cmdline_fixed_string_t show;
cmdline_fixed_string_t port;
- uint8_t port_id;
+ portid_t port_id;
cmdline_fixed_string_t rss;
cmdline_fixed_string_t reta;
uint16_t size;
@@ -2254,7 +2254,7 @@ cmdline_parse_token_string_t cmd_showport_reta_show =
cmdline_parse_token_string_t cmd_showport_reta_port =
TOKEN_STRING_INITIALIZER(struct cmd_showport_reta, port, "port");
cmdline_parse_token_num_t cmd_showport_reta_port_id =
- TOKEN_NUM_INITIALIZER(struct cmd_showport_reta, port_id, UINT8);
+ TOKEN_NUM_INITIALIZER(struct cmd_showport_reta, port_id, UINT16);
cmdline_parse_token_string_t cmd_showport_reta_rss =
TOKEN_STRING_INITIALIZER(struct cmd_showport_reta, rss, "rss");
cmdline_parse_token_string_t cmd_showport_reta_reta =
@@ -2285,7 +2285,7 @@ cmdline_parse_inst_t cmd_showport_reta = {
struct cmd_showport_rss_hash {
cmdline_fixed_string_t show;
cmdline_fixed_string
Re: [dpdk-dev] [PATCH v6 1/4] cryptodev: remove crypto vdev init API
12/10/2017 10:46, Jianfeng Tan: > --- a/doc/guides/rel_notes/deprecation.rst > +++ b/doc/guides/rel_notes/deprecation.rst > -* cryptodev: the following function is deprecated starting from 17.08 and > will > - be removed in 17.11: > - > - - ``rte_cryptodev_create_vdev`` You need to add a note in API section of release notes. If not already done, the ABIVER must be dumped.
Re: [dpdk-dev] [PATCH v2 5/5] eal/timer: honor architecture specific rdtsc hz function
12/10/2017 10:48, Bruce Richardson:
> On Wed, Oct 11, 2017 at 09:25:58PM +0200, Thomas Monjalon wrote:
> > 11/10/2017 20:57, Jerin Jacob:
> > > From: Thomas Monjalon
> > > > 22/09/2017 10:25, Gowrishankar:
> > > > > From: Jerin Jacob
> > > > >
> > > > > When calibrating the tsc frequency, first, probe the architecture
> > > > > specific
> > > > > rdtsc hz function. if not available, use the existing calibrate scheme
> > > > > to calibrate the tsc frequency.
> > > > >
> > > > > Signed-off-by: Jerin Jacob
> > > >
> > > > I agree on the idea.
> > >
> > > OK
> > >
> > > > The namespace of cycles related function in DPDK is a real mess.
> > >
> > > Absolutely!!
> > >
> > > > I think we can choose better names in this series as a first step
> > > > to tidy this mess.
> > > > I will explain below.
> > > >
> > > > At first, we should avoid TSC and RDTSC which are Intel-only wording.
> > > > The generic word could be "cycles" (the word used in arch headers),
> > > > or "ticks".
> > > > We should also name the timer sources or their function in a generic
> > > > way.
> > > > Examples: CPU cycles? fast counter? precise counter?
> > > >
> > > > Sometimes we use "hz", sometimes "freq".
> > > > It would better to keep one of them.
> > > >
> > > > > --- a/lib/librte_eal/common/eal_common_timer.c
> > > > > +++ b/lib/librte_eal/common/eal_common_timer.c
> > > > > @@ -80,8 +80,11 @@
> > > > > void
> > > > > set_tsc_freq(void)
> > > > > {
> > > > > - uint64_t freq = get_tsc_freq();
> > > > > + uint64_t freq;
> > > > >
> > > > > + freq = rte_rdtsc_arch_hz();
> > > >
> > > > This new function is arch-specific and exported as a new API.
> > >
> > > I thought of avoid exporting it. But then if the function is in
> > > lib/librte_eal/common/include/arch/../rte_cycles.h it is anyway exposed to
> > > application. i.e whatever files in lib/librte_eal/common/include/arch/../
> > > anyway exposed to application.
> >
> > Ah yes, you are right!
> >
> > > See last comment.
> > >
> > > >
> > > > > + if (!freq)
> > > > > + freq = get_tsc_freq();
> > > >
> > > > The function get_tsc_freq is guessing the freq with OS-specific method.
> > > >
> > > > > if (!freq)
> > > > > freq = estimate_tsc_freq();
> > > >
> > > > The function estimate_tsc_freq is doing an estimation based on sleep().
> > > >
> > > > At the end, the most accurate frequency is saved in
> > > > eal_tsc_resolution_hz
> > > > and can be retrieved with rte_get_tsc_hz().
> > > > I don't understand why rte_rdtsc_arch_hz() is also exported to the apps.
> > > >
> > > > TSC and HPET timer sources are wrapped in rte_get_timer_hz() in the
> > > > Similarly we can get the current timer with rte_get_timer_cycles().
> > > > In the case of TSC, it calls rte_get_tsc_cycles() which is an alias
> > > > of rte_rdtsc().
> > > > Some code is still using directly rte_rdtsc().
> > > > There is also rte_rdtsc_precise which adds a memory barrier.
> > > >
> > > > The real question is what is the right abstraction for the application?
> > > > Do we want the fastest timer? the CPU timer? a precise timer?
> > > >
> > > > I would like to see a real discussion on this topic, in order of
> > > > building
> > > > a new timer API which would alias the old one for some time.
> > >
> > > I guess, we may need to see to how abstract vmware TSC support also in
> > > proper way
> >
> > Yes
> >
> > > > If you don't want to bother with all these questions, I suggest to not
> > > > export the new function rte_rdtsc_arch_hz() and rename it to
> > > > tsc_arch_hz.
> > >
> > > If I understand it correctly, You would like to create a header file
> > > in lib/librte_eal/common/include/arch/../ which should not be exported
> > > and change
> > > the name to tsc_arch_hz.
> >
> > I had not think about the way to do this.
> > What about having internal headers in lib/librte_eal/common/arch/ ?
> >
>
> Yes, this area needs cleanup, but I also think that this patchset (and
> follow-on patch for x86-specific implementation) does not make things
> significantly worse than they are now, while also giving significant
> benefits for users both in terms of improved clock accuracy and reduced
> startup time (due to not needing sleep). Therefore I'd like to see this
> merged into 17.11 as a definite improvement, even if it does not "fix" the
> bigger issues of poor naming etc. I think this is important enough an
> improvement that we need to see it in the LTS release.
I agree Bruce.
My initial point was to avoid exporting a new function with a wrong name.
It would be better to find a way to avoid this export.
If it cannot be done in 17.11 timeframe, we can at least add the
experimental tag.
Re: [dpdk-dev] [PATCH] net/i40e: fix flexible payload configuration
> -Original Message-
> From: Chilikin, Andrey
> Sent: Saturday, October 7, 2017 2:11 AM
> To: dev@dpdk.org
> Cc: Xing, Beilei ; Wu, Jingjing
> ; Chilikin, Andrey
> Subject: [PATCH] net/i40e: fix flexible payload configuration
>
> Removed legacy writes to ORT/PIT registers from i40e_GLQF_reg_init(struct
> i40e_hw *hw) function.
> Latest NVM versions contain all relevant values and these values should not
> be overwritten by SW to maintain driver/firmware compatibility and to avoid
> conflicts with dynamic device personalization profiles.
>
> Signed-off-by: Andrey Chilikin
> ---
> drivers/net/i40e/i40e_ethdev.c | 23 +++
> 1 file changed, 11 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/net/i40e/i40e_ethdev.c b/drivers/net/i40e/i40e_ethdev.c
> index e2af51c89..d770fc343 100644
> --- a/drivers/net/i40e/i40e_ethdev.c
> +++ b/drivers/net/i40e/i40e_ethdev.c
> @@ -704,23 +704,22 @@ RTE_PMD_REGISTER_KMOD_DEP(net_i40e, "*
> igb_uio | uio_pci_generic | vfio-pci"); static inline void
> i40e_GLQF_reg_init(struct i40e_hw *hw) {
> /*
> - * Initialize registers for flexible payload, which should be set by
> NVM.
> - * This should be removed from code once it is fixed in NVM.
> + * Force global configuration for flexible payload
> + * to the first 16 bytes of the corresponding L2/L3/L4 paylod.
> + * This should be removed from code once proper
> + * configuration API is added to avoid configuration conflicts
> + * between ports of the same device.
>*/
> - I40E_WRITE_REG(hw, I40E_GLQF_ORT(18), 0x0030);
> - I40E_WRITE_REG(hw, I40E_GLQF_ORT(19), 0x0030);
> - I40E_WRITE_REG(hw, I40E_GLQF_ORT(26), 0x002B);
> - I40E_WRITE_REG(hw, I40E_GLQF_ORT(30), 0x002B);
> I40E_WRITE_REG(hw, I40E_GLQF_ORT(33), 0x00E0);
> I40E_WRITE_REG(hw, I40E_GLQF_ORT(34), 0x00E3);
> I40E_WRITE_REG(hw, I40E_GLQF_ORT(35), 0x00E6);
> - I40E_WRITE_REG(hw, I40E_GLQF_ORT(20), 0x0031);
> - I40E_WRITE_REG(hw, I40E_GLQF_ORT(23), 0x0031);
> - I40E_WRITE_REG(hw, I40E_GLQF_ORT(63), 0x002D);
> - I40E_WRITE_REG(hw, I40E_GLQF_PIT(16), 0x7480);
> - I40E_WRITE_REG(hw, I40E_GLQF_PIT(17), 0x7440);
>
> - /* Initialize registers for parsing packet type of QinQ */
> + /*
> + * Initialize registers for parsing packet type of QinQ
> + * This should be removed from code once proper
> + * configuration API is added to avoid configuration conflicts
> + * between ports of the same device.
> + */
> I40E_WRITE_REG(hw, I40E_GLQF_ORT(40), 0x0029);
> I40E_WRITE_REG(hw, I40E_GLQF_PIT(9), 0x9420); }
> --
> 2.13.0
Acked-by: Beilei Xing
Re: [dpdk-dev] [PATCH v2 5/5] eal/timer: honor architecture specific rdtsc hz function
-Original Message-
> Date: Wed, 11 Oct 2017 21:25:58 +0200
> From: Thomas Monjalon
> To: Jerin Jacob
> Cc: Gowrishankar , dev@dpdk.org, Chao
> Zhu , Bruce Richardson
> , Konstantin Ananyev
> , vikto...@rehivetech.com,
> jianbo@linaro.org
> Subject: Re: [dpdk-dev] [PATCH v2 5/5] eal/timer: honor architecture
> specific rdtsc hz function
>
> 11/10/2017 20:57, Jerin Jacob:
> > From: Thomas Monjalon
> > > 22/09/2017 10:25, Gowrishankar:
> > > > From: Jerin Jacob
> > > >
> > > > When calibrating the tsc frequency, first, probe the architecture
> > > > specific
> > > > rdtsc hz function. if not available, use the existing calibrate scheme
> > > > to calibrate the tsc frequency.
> > > >
> > > > Signed-off-by: Jerin Jacob
> > >
> > > I agree on the idea.
> >
> > OK
> >
> > > The namespace of cycles related function in DPDK is a real mess.
> >
> > Absolutely!!
> >
> > > I think we can choose better names in this series as a first step
> > > to tidy this mess.
> > > I will explain below.
> > >
> > > At first, we should avoid TSC and RDTSC which are Intel-only wording.
> > > The generic word could be "cycles" (the word used in arch headers),
> > > or "ticks".
> > > We should also name the timer sources or their function in a generic way.
> > > Examples: CPU cycles? fast counter? precise counter?
> > >
> > > Sometimes we use "hz", sometimes "freq".
> > > It would better to keep one of them.
> > >
> > > > --- a/lib/librte_eal/common/eal_common_timer.c
> > > > +++ b/lib/librte_eal/common/eal_common_timer.c
> > > > @@ -80,8 +80,11 @@
> > > > void
> > > > set_tsc_freq(void)
> > > > {
> > > > - uint64_t freq = get_tsc_freq();
> > > > + uint64_t freq;
> > > >
> > > > + freq = rte_rdtsc_arch_hz();
> > >
> > > This new function is arch-specific and exported as a new API.
> >
> > I thought of avoid exporting it. But then if the function is in
> > lib/librte_eal/common/include/arch/../rte_cycles.h it is anyway exposed to
> > application. i.e whatever files in lib/librte_eal/common/include/arch/../
> > anyway exposed to application.
>
> Ah yes, you are right!
>
> > See last comment.
> >
> > >
> > > > + if (!freq)
> > > > + freq = get_tsc_freq();
> > >
> > > The function get_tsc_freq is guessing the freq with OS-specific method.
> > >
> > > > if (!freq)
> > > > freq = estimate_tsc_freq();
> > >
> > > The function estimate_tsc_freq is doing an estimation based on sleep().
> > >
> > > At the end, the most accurate frequency is saved in eal_tsc_resolution_hz
> > > and can be retrieved with rte_get_tsc_hz().
> > > I don't understand why rte_rdtsc_arch_hz() is also exported to the apps.
> > >
> > > TSC and HPET timer sources are wrapped in rte_get_timer_hz() in the
> > > Similarly we can get the current timer with rte_get_timer_cycles().
> > > In the case of TSC, it calls rte_get_tsc_cycles() which is an alias
> > > of rte_rdtsc().
> > > Some code is still using directly rte_rdtsc().
> > > There is also rte_rdtsc_precise which adds a memory barrier.
> > >
> > > The real question is what is the right abstraction for the application?
> > > Do we want the fastest timer? the CPU timer? a precise timer?
> > >
> > > I would like to see a real discussion on this topic, in order of building
> > > a new timer API which would alias the old one for some time.
> >
> > I guess, we may need to see to how abstract vmware TSC support also in
> > proper way
>
> Yes
>
> > > If you don't want to bother with all these questions, I suggest to not
> > > export the new function rte_rdtsc_arch_hz() and rename it to tsc_arch_hz.
> >
> > If I understand it correctly, You would like to create a header file
> > in lib/librte_eal/common/include/arch/../ which should not be exported and
> > change
> > the name to tsc_arch_hz.
>
> I had not think about the way to do this.
> What about having internal headers in lib/librte_eal/common/arch/ ?
Looks bit odd when compare with existing scheme of arch function management.
Anyway the whole timer stuff needs to be cleanup.
I think, we can take that when do the cleanup.
On the other side, This API has real ARM64, PPC, x86 implementation now.
I think,it is important to have this feature for v17.11
>
Re: [dpdk-dev] [PATCH v2 5/5] eal/timer: honor architecture specific rdtsc hz function
On Thu, Oct 12, 2017 at 12:12:48PM +0200, Thomas Monjalon wrote:
> 12/10/2017 10:48, Bruce Richardson:
> > On Wed, Oct 11, 2017 at 09:25:58PM +0200, Thomas Monjalon wrote:
> > > 11/10/2017 20:57, Jerin Jacob:
> > > > From: Thomas Monjalon
> > > > > 22/09/2017 10:25, Gowrishankar:
> > > > > > From: Jerin Jacob
> > > > > >
> > > > > > When calibrating the tsc frequency, first, probe the architecture
> > > > > > specific
> > > > > > rdtsc hz function. if not available, use the existing calibrate
> > > > > > scheme
> > > > > > to calibrate the tsc frequency.
> > > > > >
> > > > > > Signed-off-by: Jerin Jacob
> > > > >
> > > > > I agree on the idea.
> > > >
> > > > OK
> > > >
> > > > > The namespace of cycles related function in DPDK is a real mess.
> > > >
> > > > Absolutely!!
> > > >
> > > > > I think we can choose better names in this series as a first step
> > > > > to tidy this mess.
> > > > > I will explain below.
> > > > >
> > > > > At first, we should avoid TSC and RDTSC which are Intel-only wording.
> > > > > The generic word could be "cycles" (the word used in arch headers),
> > > > > or "ticks".
> > > > > We should also name the timer sources or their function in a generic
> > > > > way.
> > > > > Examples: CPU cycles? fast counter? precise counter?
> > > > >
> > > > > Sometimes we use "hz", sometimes "freq".
> > > > > It would better to keep one of them.
> > > > >
> > > > > > --- a/lib/librte_eal/common/eal_common_timer.c
> > > > > > +++ b/lib/librte_eal/common/eal_common_timer.c
> > > > > > @@ -80,8 +80,11 @@
> > > > > > void
> > > > > > set_tsc_freq(void)
> > > > > > {
> > > > > > - uint64_t freq = get_tsc_freq();
> > > > > > + uint64_t freq;
> > > > > >
> > > > > > + freq = rte_rdtsc_arch_hz();
> > > > >
> > > > > This new function is arch-specific and exported as a new API.
> > > >
> > > > I thought of avoid exporting it. But then if the function is in
> > > > lib/librte_eal/common/include/arch/../rte_cycles.h it is anyway exposed
> > > > to
> > > > application. i.e whatever files in
> > > > lib/librte_eal/common/include/arch/../
> > > > anyway exposed to application.
> > >
> > > Ah yes, you are right!
> > >
> > > > See last comment.
> > > >
> > > > >
> > > > > > + if (!freq)
> > > > > > + freq = get_tsc_freq();
> > > > >
> > > > > The function get_tsc_freq is guessing the freq with OS-specific
> > > > > method.
> > > > >
> > > > > > if (!freq)
> > > > > > freq = estimate_tsc_freq();
> > > > >
> > > > > The function estimate_tsc_freq is doing an estimation based on
> > > > > sleep().
> > > > >
> > > > > At the end, the most accurate frequency is saved in
> > > > > eal_tsc_resolution_hz
> > > > > and can be retrieved with rte_get_tsc_hz().
> > > > > I don't understand why rte_rdtsc_arch_hz() is also exported to the
> > > > > apps.
> > > > >
> > > > > TSC and HPET timer sources are wrapped in rte_get_timer_hz() in the
> > > > > Similarly we can get the current timer with rte_get_timer_cycles().
> > > > > In the case of TSC, it calls rte_get_tsc_cycles() which is an alias
> > > > > of rte_rdtsc().
> > > > > Some code is still using directly rte_rdtsc().
> > > > > There is also rte_rdtsc_precise which adds a memory barrier.
> > > > >
> > > > > The real question is what is the right abstraction for the
> > > > > application?
> > > > > Do we want the fastest timer? the CPU timer? a precise timer?
> > > > >
> > > > > I would like to see a real discussion on this topic, in order of
> > > > > building
> > > > > a new timer API which would alias the old one for some time.
> > > >
> > > > I guess, we may need to see to how abstract vmware TSC support also in
> > > > proper way
> > >
> > > Yes
> > >
> > > > > If you don't want to bother with all these questions, I suggest to not
> > > > > export the new function rte_rdtsc_arch_hz() and rename it to
> > > > > tsc_arch_hz.
> > > >
> > > > If I understand it correctly, You would like to create a header file
> > > > in lib/librte_eal/common/include/arch/../ which should not be exported
> > > > and change
> > > > the name to tsc_arch_hz.
> > >
> > > I had not think about the way to do this.
> > > What about having internal headers in lib/librte_eal/common/arch/ ?
> > >
> >
> > Yes, this area needs cleanup, but I also think that this patchset (and
> > follow-on patch for x86-specific implementation) does not make things
> > significantly worse than they are now, while also giving significant
> > benefits for users both in terms of improved clock accuracy and reduced
> > startup time (due to not needing sleep). Therefore I'd like to see this
> > merged into 17.11 as a definite improvement, even if it does not "fix" the
> > bigger issues of poor naming etc. I think this is important enough an
> > improvement that we need to see it in the LTS release.
>
> I agree Bruce.
> My initial point was to avoid exporting a new function with a wrong name.
> It would be better to find a way t
Re: [dpdk-dev] [PATCH] examples/ip_reassembly: use pktmbuf to create pool
> -Original Message-
> From: Hemant Agrawal [mailto:hemant.agra...@nxp.com]
> Sent: Wednesday, September 6, 2017 10:34 AM
> To: dev@dpdk.org
> Cc: Ananyev, Konstantin ; ashish.j...@nxp.com
> Subject: [PATCH] examples/ip_reassembly: use pktmbuf to create pool
>
> From: Ashish Jain
>
> Use of rte_mempool_create() with the helper provided in
> librte_mbuf: rte_pktmbuf_pool_create().
> This is the preferred way to create a mbuf pool else
> it may not work on implementation using the HW buffer pool
>
> Signed-off-by: Ashish Jain
> ---
> examples/ip_reassembly/main.c | 13 ++---
> 1 file changed, 6 insertions(+), 7 deletions(-)
>
> diff --git a/examples/ip_reassembly/main.c b/examples/ip_reassembly/main.c
> index e62636c..20caeb3 100644
> --- a/examples/ip_reassembly/main.c
> +++ b/examples/ip_reassembly/main.c
> @@ -84,8 +84,7 @@
> #define MAX_JUMBO_PKT_LEN 9600
>
> #define BUF_SIZERTE_MBUF_DEFAULT_DATAROOM
> -#define MBUF_SIZE\
> - (BUF_SIZE + sizeof(struct rte_mbuf) + RTE_PKTMBUF_HEADROOM)
> +#define MBUF_DATA_SIZE RTE_MBUF_DEFAULT_BUF_SIZE
>
> #define NB_MBUF 8192
>
> @@ -909,11 +908,11 @@ setup_queue_tbl(struct rx_queue *rxq, uint32_t lcore,
> uint32_t queue)
>
> snprintf(buf, sizeof(buf), "mbuf_pool_%u_%u", lcore, queue);
>
> - if ((rxq->pool = rte_mempool_create(buf, nb_mbuf, MBUF_SIZE, 0,
> - sizeof(struct rte_pktmbuf_pool_private),
> - rte_pktmbuf_pool_init, NULL, rte_pktmbuf_init, NULL,
> - socket, MEMPOOL_F_SP_PUT | MEMPOOL_F_SC_GET)) == NULL) {
> - RTE_LOG(ERR, IP_RSMBL, "mempool_create(%s) failed", buf);
> + rxq->pool = rte_pktmbuf_pool_create(buf, nb_mbuf, 0, 0,
> + MBUF_DATA_SIZE, socket);
As we can't pass SC|SP anymore can we then setup mempool cache size to some
non-zero value?
Konstantin
> + if (rxq->pool == NULL) {
> + RTE_LOG(ERR, IP_RSMBL,
> + "rte_pktmbuf_pool_create(%s) failed", buf);
> return -1;
> }
>
> --
> 2.7.4
[dpdk-dev] [PATCH v5 00/20] Move PCI away from the EAL
Hi all, Here is a new version of the PCI bus move out of the EAL. The EAL PCI implementation is divided in two parts: - librte_pci: library offering helpers to handle PCI objects - librte_bus_pci: bus driver for PCI devices This allows other libraries / tools to use PCI elements (location, mappings, parsing operations, etc) without forcing a dependency on a bus driver. The latter should not have to export helpers that others might need. It is focused on defining the rte_pci_device, rte_pci_driver objects and their handling. The cryptodev library has hard dependencies on rte_pci_devices (used by generic probe function). Other similar libs (ether and eventdev) avoided the issue by inlining such functions and expecting users to include the relevant headers once the PCI bus has already been built. After review from Declan, he proposed to submit a patch removing this dependency. Once this patch is submitted, the relevant commit will be dropped from this patchset. v2: + Made rte_eal_using_phys_addrs common to both linux and bsd interfaces. + Added documentation of EAL API changes in release note. + Fixed a few rebase-related mistakes. + Fixed parallel build race condition reported by Luca Boccassi. + Grouped together commits breaking compilation: -> pci: introduce PCI lib and bus -> lib: include rte_bus_pci -> drivers: include rte_bus_pci -> test: include rte_bus_pci -> app/testpmd: include rte_bus_pci -> cryptodev: move PCI specific helpers to drivers/crypto Until all of them have been applied, compilation is broken. I am currently wondering whether merging some of them might be sensible. + Not included in this series: Several filesystem-related functions are currently private to the EAL and directly linked. This is not good, but the solution seems to be to have a new lib offering an FS abstraction. This seems an overreach for this patchset and should probably come in a second step. v3: + Fixed .map versioning + merged one commit breaking the build into the main commit moving code around. Other such commits are still present, as they only break specific subsystems (lib, drivers, apps, cryptodev). Merging them all within the one main commit does not seem right. As such, build is still broken from * pci: introduce PCI lib and bus until * cryptodev: move PCI specific helpers to drivers/crypto v4: + Rebased unto master, with new PCI functionalities integrated. + Removed the exposition of private EAL functions. While one commit did deal with this for one function, the issue is more widespread and should be fixed in a more generic way. + Introduced new PCI address parsing function, deprecating the old ones. + Fix conflict with bonding PMD regarding pci_addr_cmp function name. v5: + Rebase unto master: fix a few compilation issues with the header change. + Make more PCI bus functions private. ~*~ Compilation is still broken from * pci: introduce PCI lib and bus until * net/bonding: use local prefix for local function Gaetan Rivet (20): eal: include debug header in bus source ethdev: remove useless PCI dependency pci: introduce PCI lib and bus lib: include PCI bus header drivers: include PCI bus header test: include PCI bus header app/testpmd: include PCI bus header cryptodev: move PCI specific helpers to drivers/crypto net/bonding: use local prefix for local function pci: avoid inlining functions pci: avoid over-complicated macro pci: deprecate misnamed functions pci: introduce PCI address parsing function pci: make specialized parsing functions private bus/pci: use new PCI addr parsing function bus/pci: do not expose private functions bus/pci: do not expose PCI match function bus/pci: do not expose IOVA mode getter doc: add notes on EAL PCI API update bus: rename scan policy as probe policy app/test-pmd/testpmd.h | 1 + config/common_base | 10 + doc/guides/rel_notes/deprecation.rst | 10 + doc/guides/rel_notes/release_17_11.rst | 28 + drivers/bus/Makefile | 2 + drivers/bus/pci/Makefile | 59 ++ drivers/bus/pci/bsd/Makefile | 32 + drivers/bus/pci/bsd/pci.c| 681 +++ drivers/bus/pci/include/rte_bus_pci.h| 350 ++ drivers/bus/pci/linux/Makefile | 37 + drivers/bus/pci/linux/pci.c | 821 +++ drivers/bus/pci/linux/pci_init.h | 97 +++ drivers/bus/pci/linux/pci_uio.c | 568 drivers/bus/pci/linux/pci_vfio.c | 757 + drivers/bus/pci/linux/vfio_mp_sync.c | 426 drivers/bus/pci/pci_common.c | 5
[dpdk-dev] [PATCH v5 02/20] ethdev: remove useless PCI dependency
Signed-off-by: Gaetan Rivet --- lib/librte_ether/rte_ethdev.c | 1 - 1 file changed, 1 deletion(-) diff --git a/lib/librte_ether/rte_ethdev.c b/lib/librte_ether/rte_ethdev.c index 9460161..bb08204 100644 --- a/lib/librte_ether/rte_ethdev.c +++ b/lib/librte_ether/rte_ethdev.c @@ -47,7 +47,6 @@ #include #include #include -#include #include #include #include -- 2.1.4
[dpdk-dev] [PATCH v5 05/20] drivers: include PCI bus header
Devices and drivers are now defined within the bus-specific PCI header. Update drivers. Signed-off-by: Gaetan Rivet --- drivers/bus/pci/bsd/pci.c | 1 + drivers/bus/pci/linux/pci.c| 1 + drivers/bus/pci/linux/pci_uio.c| 1 + drivers/bus/pci/linux/pci_vfio.c | 1 + drivers/bus/pci/linux/vfio_mp_sync.c | 1 + drivers/bus/pci/pci_common.c | 1 + drivers/bus/pci/pci_common_uio.c | 1 + drivers/bus/pci/private.h | 1 + drivers/crypto/qat/qat_qp.c| 1 + drivers/event/octeontx/ssovf_probe.c | 1 + drivers/mempool/octeontx/octeontx_fpavf.c | 2 +- drivers/net/ark/ark_ethdev.c | 1 + drivers/net/avp/avp_ethdev.c | 2 ++ drivers/net/bnxt/bnxt.h| 1 + drivers/net/bonding/rte_eth_bond_args.c| 1 + drivers/net/cxgbe/base/adapter.h | 1 + drivers/net/cxgbe/cxgbe_ethdev.c | 1 + drivers/net/e1000/em_ethdev.c | 1 + drivers/net/e1000/igb_ethdev.c | 1 + drivers/net/e1000/igb_pf.c | 1 + drivers/net/ena/ena_ethdev.h | 1 + drivers/net/enic/base/vnic_dev.h | 4 +++- drivers/net/enic/enic_ethdev.c | 1 + drivers/net/enic/enic_main.c | 1 + drivers/net/i40e/i40e_ethdev.c | 1 + drivers/net/i40e/i40e_ethdev_vf.c | 1 + drivers/net/ixgbe/ixgbe_ethdev.c | 1 + drivers/net/ixgbe/ixgbe_ethdev.h | 1 + drivers/net/mlx4/mlx4_ethdev.c | 2 +- drivers/net/mlx5/mlx5.c| 1 + drivers/net/mlx5/mlx5_ethdev.c | 1 + drivers/net/nfp/nfp_nfpu.c | 2 +- drivers/net/nfp/nfp_nfpu.h | 2 +- drivers/net/octeontx/base/octeontx_pkivf.c | 2 +- drivers/net/octeontx/base/octeontx_pkovf.c | 2 +- drivers/net/sfc/sfc.h | 1 + drivers/net/sfc/sfc_ethdev.c | 1 + drivers/net/thunderx/nicvf_ethdev.c| 1 + drivers/net/virtio/virtio_ethdev.c | 1 + drivers/net/virtio/virtio_pci.h| 1 + drivers/net/vmxnet3/vmxnet3_ethdev.c | 1 + 41 files changed, 44 insertions(+), 7 deletions(-) diff --git a/drivers/bus/pci/bsd/pci.c b/drivers/bus/pci/bsd/pci.c index 599519a..753d914 100644 --- a/drivers/bus/pci/bsd/pci.c +++ b/drivers/bus/pci/bsd/pci.c @@ -57,6 +57,7 @@ #include #include #include +#include #include #include #include diff --git a/drivers/bus/pci/linux/pci.c b/drivers/bus/pci/linux/pci.c index c1fcc56..422579f 100644 --- a/drivers/bus/pci/linux/pci.c +++ b/drivers/bus/pci/linux/pci.c @@ -37,6 +37,7 @@ #include #include #include +#include #include #include #include diff --git a/drivers/bus/pci/linux/pci_uio.c b/drivers/bus/pci/linux/pci_uio.c index d3f0556..940fa21 100644 --- a/drivers/bus/pci/linux/pci_uio.c +++ b/drivers/bus/pci/linux/pci_uio.c @@ -47,6 +47,7 @@ #include #include +#include #include #include #include diff --git a/drivers/bus/pci/linux/pci_vfio.c b/drivers/bus/pci/linux/pci_vfio.c index e69e4d3..855e2b9 100644 --- a/drivers/bus/pci/linux/pci_vfio.c +++ b/drivers/bus/pci/linux/pci_vfio.c @@ -42,6 +42,7 @@ #include #include +#include #include #include diff --git a/drivers/bus/pci/linux/vfio_mp_sync.c b/drivers/bus/pci/linux/vfio_mp_sync.c index dd24e86..d07687f 100644 --- a/drivers/bus/pci/linux/vfio_mp_sync.c +++ b/drivers/bus/pci/linux/vfio_mp_sync.c @@ -50,6 +50,7 @@ #include #include +#include #include #include diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c index 7b68e99..cd95d1a 100644 --- a/drivers/bus/pci/pci_common.c +++ b/drivers/bus/pci/pci_common.c @@ -45,6 +45,7 @@ #include #include #include +#include #include #include #include diff --git a/drivers/bus/pci/pci_common_uio.c b/drivers/bus/pci/pci_common_uio.c index 4365660..544c606 100644 --- a/drivers/bus/pci/pci_common_uio.c +++ b/drivers/bus/pci/pci_common_uio.c @@ -40,6 +40,7 @@ #include #include +#include #include #include #include diff --git a/drivers/bus/pci/private.h b/drivers/bus/pci/private.h index 7ff1fc4..fdc2c81 100644 --- a/drivers/bus/pci/private.h +++ b/drivers/bus/pci/private.h @@ -37,6 +37,7 @@ #include #include #include +#include struct rte_pci_driver; struct rte_pci_device; diff --git a/drivers/crypto/qat/qat_qp.c b/drivers/crypto/qat/qat_qp.c index 34f75ca..8489662 100644 --- a/drivers/crypto/qat/qat_qp.c +++ b/drivers/crypto/qat/qat_qp.c @@ -37,6 +37,7 @@ #include #include #include +#include #include #include diff --git a/drivers/event/octeontx/ssovf_probe.c b/drivers/event/octeontx/ssovf_probe.c index e1c0c6d..1cac4bc 100644 --- a/drivers/event/octeontx/ssovf_probe.c +++ b/drivers/event/octeontx/ssovf_probe.c @@ -35,6 +35,7 @@ #include #include #include +#include #include "ssovf_evdev.h" diff --git a/drivers/mempool/octeontx/octeontx_fpavf.c
[dpdk-dev] [PATCH v5 01/20] eal: include debug header in bus source
This header is included through rte_pci.h, which will be removed once the PCI bus is moved out of the EAL. Signed-off-by: Gaetan Rivet --- lib/librte_eal/common/eal_common_bus.c | 1 + 1 file changed, 1 insertion(+) diff --git a/lib/librte_eal/common/eal_common_bus.c b/lib/librte_eal/common/eal_common_bus.c index c3c77f4..3c66a02 100644 --- a/lib/librte_eal/common/eal_common_bus.c +++ b/lib/librte_eal/common/eal_common_bus.c @@ -35,6 +35,7 @@ #include #include +#include #include "eal_private.h" -- 2.1.4
[dpdk-dev] [PATCH v5 04/20] lib: include PCI bus header
Devices and drivers are now defined within the bus-specific PCI header.
Update the libraries, as structuraly unsound as it may be.
Signed-off-by: Gaetan Rivet
---
lib/librte_ether/rte_ethdev_pci.h | 1 +
lib/librte_eventdev/rte_eventdev_pmd_pci.h | 1 +
2 files changed, 2 insertions(+)
diff --git a/lib/librte_ether/rte_ethdev_pci.h
b/lib/librte_ether/rte_ethdev_pci.h
index 56b1072..722075e 100644
--- a/lib/librte_ether/rte_ethdev_pci.h
+++ b/lib/librte_ether/rte_ethdev_pci.h
@@ -36,6 +36,7 @@
#include
#include
+#include
#include
/**
diff --git a/lib/librte_eventdev/rte_eventdev_pmd_pci.h
b/lib/librte_eventdev/rte_eventdev_pmd_pci.h
index b6bd731..ade32b5 100644
--- a/lib/librte_eventdev/rte_eventdev_pmd_pci.h
+++ b/lib/librte_eventdev/rte_eventdev_pmd_pci.h
@@ -50,6 +50,7 @@ extern "C" {
#include
#include
#include
+#include
#include "rte_eventdev_pmd.h"
--
2.1.4
[dpdk-dev] [PATCH v5 09/20] net/bonding: use local prefix for local function
Signed-off-by: Gaetan Rivet
---
drivers/net/bonding/rte_eth_bond_args.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/bonding/rte_eth_bond_args.c
b/drivers/net/bonding/rte_eth_bond_args.c
index dea0265..e816da3 100644
--- a/drivers/net/bonding/rte_eth_bond_args.c
+++ b/drivers/net/bonding/rte_eth_bond_args.c
@@ -90,7 +90,7 @@ find_port_id_by_dev_name(const char *name)
}
static inline int
-pci_addr_cmp(const struct rte_device *dev, const void *_pci_addr)
+bond_pci_addr_cmp(const struct rte_device *dev, const void *_pci_addr)
{
struct rte_pci_device *pdev;
const struct rte_pci_addr *paddr = _pci_addr;
@@ -119,7 +119,7 @@ parse_port_id(const char *port_str)
/* try parsing as pci address, physical devices */
if (pci_bus->parse(port_str, &dev_addr) == 0) {
- dev = pci_bus->find_device(NULL, pci_addr_cmp, &dev_addr);
+ dev = pci_bus->find_device(NULL, bond_pci_addr_cmp, &dev_addr);
if (dev == NULL) {
RTE_LOG(ERR, PMD, "unable to find PCI device\n");
return -1;
--
2.1.4
[dpdk-dev] [PATCH v5 06/20] test: include PCI bus header
Devices and drivers are now defined within the bus-specific PCI header. Update test applications. Signed-off-by: Gaetan Rivet --- test/test/test_kni.c| 1 + test/test/virtual_pmd.c | 1 + 2 files changed, 2 insertions(+) diff --git a/test/test/test_kni.c b/test/test/test_kni.c index 2450c9f..b956727 100644 --- a/test/test/test_kni.c +++ b/test/test/test_kni.c @@ -42,6 +42,7 @@ #include #include #include +#include #include #include diff --git a/test/test/virtual_pmd.c b/test/test/virtual_pmd.c index 09daf6c..b57a949 100644 --- a/test/test/virtual_pmd.c +++ b/test/test/virtual_pmd.c @@ -34,6 +34,7 @@ #include #include #include +#include #include #include #include -- 2.1.4
[dpdk-dev] [PATCH v5 07/20] app/testpmd: include PCI bus header
Devices and drivers are now defined within the bus-specific PCI header. Update applications. Signed-off-by: Gaetan Rivet --- app/test-pmd/testpmd.h | 1 + 1 file changed, 1 insertion(+) diff --git a/app/test-pmd/testpmd.h b/app/test-pmd/testpmd.h index 0b8e624..b05c8a0 100644 --- a/app/test-pmd/testpmd.h +++ b/app/test-pmd/testpmd.h @@ -35,6 +35,7 @@ #define _TESTPMD_H_ #include +#include #include #include -- 2.1.4
[dpdk-dev] [PATCH v5 08/20] cryptodev: move PCI specific helpers to drivers/crypto
Those helpers rely on the PCI bus driver implementation. Other similar libraries relied on the bus-specifics being handled in inlined functions, to be compiled on demand by drivers, once the proper PCI dependency has been settled. This seems unsafe. Move the PCI-specific helpers out of the lib directory to the drivers/crypto directory, properly following the dependency hierarchy. Signed-off-by: Gaetan Rivet --- drivers/crypto/Makefile | 4 +- drivers/crypto/pci/Makefile | 52 + drivers/crypto/pci/rte_cryptodev_pci.c | 128 +++ drivers/crypto/pci/rte_cryptodev_pci.h | 94 + drivers/crypto/pci/rte_cryptodev_pci_version.map | 7 ++ lib/librte_cryptodev/Makefile| 1 - lib/librte_cryptodev/rte_cryptodev_pci.h | 92 lib/librte_cryptodev/rte_cryptodev_pmd.c | 94 - lib/librte_cryptodev/rte_cryptodev_version.map | 2 - 9 files changed, 284 insertions(+), 190 deletions(-) create mode 100644 drivers/crypto/pci/Makefile create mode 100644 drivers/crypto/pci/rte_cryptodev_pci.c create mode 100644 drivers/crypto/pci/rte_cryptodev_pci.h create mode 100644 drivers/crypto/pci/rte_cryptodev_pci_version.map delete mode 100644 lib/librte_cryptodev/rte_cryptodev_pci.h diff --git a/drivers/crypto/Makefile b/drivers/crypto/Makefile index 7a719b9..cfd6cb6 100644 --- a/drivers/crypto/Makefile +++ b/drivers/crypto/Makefile @@ -33,6 +33,8 @@ include $(RTE_SDK)/mk/rte.vars.mk core-libs := librte_eal librte_mbuf librte_mempool librte_ring librte_cryptodev +DIRS-$(CONFIG_RTE_LIBRTE_PCI_BUS) += pci +DEPDIRS-pci = $(core-libs) DIRS-$(CONFIG_RTE_LIBRTE_PMD_AESNI_GCM) += aesni_gcm DEPDIRS-aesni_gcm = $(core-libs) DIRS-$(CONFIG_RTE_LIBRTE_PMD_AESNI_MB) += aesni_mb @@ -42,7 +44,7 @@ DEPDIRS-armv8 = $(core-libs) DIRS-$(CONFIG_RTE_LIBRTE_PMD_OPENSSL) += openssl DEPDIRS-openssl = $(core-libs) DIRS-$(CONFIG_RTE_LIBRTE_PMD_QAT) += qat -DEPDIRS-qat = $(core-libs) +DEPDIRS-qat = $(core-libs) librte_cryptodev_pci DIRS-$(CONFIG_RTE_LIBRTE_PMD_CRYPTO_SCHEDULER) += scheduler DEPDIRS-scheduler = $(core-libs) librte_kvargs librte_reorder DIRS-$(CONFIG_RTE_LIBRTE_PMD_SNOW3G) += snow3g diff --git a/drivers/crypto/pci/Makefile b/drivers/crypto/pci/Makefile new file mode 100644 index 000..da819f2 --- /dev/null +++ b/drivers/crypto/pci/Makefile @@ -0,0 +1,52 @@ +# BSD LICENSE +# +# Copyright(c) 2017 6WIND S.A. All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions +# are met: +# +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in +# the documentation and/or other materials provided with the +# distribution. +# * Neither the name of 6WIND S.A. nor the names of its +# contributors may be used to endorse or promote products derived +# from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS +# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT +# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR +# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT +# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, +# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT +# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +include $(RTE_SDK)/mk/rte.vars.mk + +# library name +LIB = librte_cryptodev_pci.a + +# library version +LIBABIVER := 1 + +# build flags +CFLAGS += -O3 +CFLAGS += $(WERROR_FLAGS) + +# library source files +SRCS-y += rte_cryptodev_pci.c + +# export include files +SYMLINK-y-include += rte_cryptodev_pci.h + +# versioning export map +EXPORT_MAP := rte_cryptodev_pci_version.map + +include $(RTE_SDK)/mk/rte.lib.mk diff --git a/drivers/crypto/pci/rte_cryptodev_pci.c b/drivers/crypto/pci/rte_cryptodev_pci.c new file mode 100644 index 000..a2a1366 --- /dev/null +++ b/drivers/crypto/pci/rte_cryptodev_pci.c @@ -0,0 +1,128 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2017 6WIND S.A. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * + * *
[dpdk-dev] [PATCH v5 10/20] pci: avoid inlining functions
Parsing operations should not happen in performance critical sections.
Headers should not propose implementations unless duly required.
Signed-off-by: Gaetan Rivet
---
lib/librte_pci/include/rte_pci.h | 69 --
lib/librte_pci/rte_pci.c | 65 +++
lib/librte_pci/rte_pci_version.map | 4 +++
3 files changed, 75 insertions(+), 63 deletions(-)
diff --git a/lib/librte_pci/include/rte_pci.h b/lib/librte_pci/include/rte_pci.h
index 2647568..fe4d411 100644
--- a/lib/librte_pci/include/rte_pci.h
+++ b/lib/librte_pci/include/rte_pci.h
@@ -134,19 +134,6 @@ struct mapped_pci_resource {
/** mapped pci device list */
TAILQ_HEAD(mapped_pci_res_list, mapped_pci_resource);
-/**< Internal use only - Macro used by pci addr parsing functions **/
-#define GET_PCIADDR_FIELD(in, fd, lim, dlm) \
-do { \
- unsigned long val; \
- char *end; \
- errno = 0; \
- val = strtoul((in), &end, 16); \
- if (errno != 0 || end[0] != (dlm) || val > (lim)) \
- return -EINVAL; \
- (fd) = (typeof (fd))val;\
- (in) = end + 1; \
-} while(0)
-
/**
* Utility function to produce a PCI Bus-Device-Function value
* given a string representation. Assumes that the BDF is provided without
@@ -160,15 +147,7 @@ do {
\
* @return
* 0 on success, negative on error.
*/
-static inline int
-eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr)
-{
- dev_addr->domain = 0;
- GET_PCIADDR_FIELD(input, dev_addr->bus, UINT8_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->devid, UINT8_MAX, '.');
- GET_PCIADDR_FIELD(input, dev_addr->function, UINT8_MAX, 0);
- return 0;
-}
+int eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr);
/**
* Utility function to produce a PCI Bus-Device-Function value
@@ -182,16 +161,7 @@ eal_parse_pci_BDF(const char *input, struct rte_pci_addr
*dev_addr)
* @return
* 0 on success, negative on error.
*/
-static inline int
-eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr)
-{
- GET_PCIADDR_FIELD(input, dev_addr->domain, UINT16_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->bus, UINT8_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->devid, UINT8_MAX, '.');
- GET_PCIADDR_FIELD(input, dev_addr->function, UINT8_MAX, 0);
- return 0;
-}
-#undef GET_PCIADDR_FIELD
+int eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr);
/**
* Utility function to write a pci device name, this device name can later be
@@ -205,17 +175,9 @@ eal_parse_pci_DomBDF(const char *input, struct
rte_pci_addr *dev_addr)
* @param size
* The output buffer size
*/
-static inline void
-rte_pci_device_name(const struct rte_pci_addr *addr,
- char *output, size_t size)
-{
- RTE_VERIFY(size >= PCI_PRI_STR_SIZE);
- RTE_VERIFY(snprintf(output, size, PCI_PRI_FMT,
- addr->domain, addr->bus,
- addr->devid, addr->function) >= 0);
-}
+void rte_pci_device_name(const struct rte_pci_addr *addr,
+char *output, size_t size);
-/* Compare two PCI device addresses. */
/**
* Utility function to compare two PCI device addresses.
*
@@ -228,27 +190,8 @@ rte_pci_device_name(const struct rte_pci_addr *addr,
* Positive on addr is greater than addr2.
* Negative on addr is less than addr2, or error.
*/
-static inline int
-rte_eal_compare_pci_addr(const struct rte_pci_addr *addr,
-const struct rte_pci_addr *addr2)
-{
- uint64_t dev_addr, dev_addr2;
-
- if ((addr == NULL) || (addr2 == NULL))
- return -1;
-
- dev_addr = ((uint64_t)addr->domain << 24) |
- (addr->bus << 16) | (addr->devid << 8) | addr->function;
- dev_addr2 = ((uint64_t)addr2->domain << 24) |
- (addr2->bus << 16) | (addr2->devid << 8) | addr2->function;
-
- if (dev_addr > dev_addr2)
- return 1;
- else if (dev_addr < dev_addr2)
- return -1;
- else
- return 0;
-}
+int rte_eal_compare_pci_addr(const struct rte_pci_addr *addr,
+const struct rte_pci_addr *addr2);
/**
* Map a particular resource from a file.
diff --git a/lib/librte_pci/rte_pci.c b/lib/librte_pci/rte_pci.c
index 9dfdd3f..8584b55 100644
--- a/lib/librte_pci/rte_pci.c
+++ b/lib/librte_pci/rte_pci.c
@@ -53,6 +53,71 @@
#include "rte_pci.h"
+/* Macro used by pci addr parsing functions. **/
+#d
[dpdk-dev] [PATCH v5 15/20] bus/pci: use new PCI addr parsing function
Signed-off-by: Gaetan Rivet --- drivers/bus/pci/pci_common.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c index cd95d1a..d7a1c05 100644 --- a/drivers/bus/pci/pci_common.c +++ b/drivers/bus/pci/pci_common.c @@ -437,8 +437,7 @@ pci_parse(const char *name, void *addr) struct rte_pci_addr pci_addr; bool parse; - parse = (eal_parse_pci_BDF(name, &pci_addr) == 0 || -eal_parse_pci_DomBDF(name, &pci_addr) == 0); + parse = (pci_addr_parse(name, &pci_addr) == 0); if (parse && addr != NULL) *out = pci_addr; return parse == false; -- 2.1.4
[dpdk-dev] [PATCH v5 11/20] pci: avoid over-complicated macro
Using a macro helps writing the code to the detriment of the reader in
this case. This is backward. Write once, read many.
The few LOCs gained is not worth the opacity of the implementation.
Signed-off-by: Gaetan Rivet
---
lib/librte_pci/rte_pci.c | 65 ++--
1 file changed, 46 insertions(+), 19 deletions(-)
diff --git a/lib/librte_pci/rte_pci.c b/lib/librte_pci/rte_pci.c
index 8584b55..cbb5359 100644
--- a/lib/librte_pci/rte_pci.c
+++ b/lib/librte_pci/rte_pci.c
@@ -53,36 +53,63 @@
#include "rte_pci.h"
-/* Macro used by pci addr parsing functions. **/
-#define GET_PCIADDR_FIELD(in, fd, lim, dlm) \
-do {\
- unsigned long val; \
- char *end; \
- errno = 0; \
- val = strtoul((in), &end, 16); \
- if (errno != 0 || end[0] != (dlm) || val > (lim)) \
- return -EINVAL; \
- (fd) = (typeof (fd))val;\
- (in) = end + 1; \
-} while(0)
+static inline const char *
+get_u8_pciaddr_field(const char *in, void *_u8, char dlm)
+{
+ unsigned long val;
+ uint8_t *u8 = _u8;
+ char *end;
+
+ errno = 0;
+ val = strtoul(in, &end, 16);
+ if (errno != 0 || end[0] != dlm || val > UINT8_MAX) {
+ errno = errno ? errno : EINVAL;
+ return NULL;
+ }
+ *u8 = (uint8_t)val;
+ return end + 1;
+}
int
eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr)
{
+ const char *in = input;
+
dev_addr->domain = 0;
- GET_PCIADDR_FIELD(input, dev_addr->bus, UINT8_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->devid, UINT8_MAX, '.');
- GET_PCIADDR_FIELD(input, dev_addr->function, UINT8_MAX, 0);
+ in = get_u8_pciaddr_field(in, &dev_addr->bus, ':');
+ if (in == NULL)
+ return -EINVAL;
+ in = get_u8_pciaddr_field(in, &dev_addr->devid, '.');
+ if (in == NULL)
+ return -EINVAL;
+ in = get_u8_pciaddr_field(in, &dev_addr->function, '\0');
+ if (in == NULL)
+ return -EINVAL;
return 0;
}
int
eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr)
{
- GET_PCIADDR_FIELD(input, dev_addr->domain, UINT16_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->bus, UINT8_MAX, ':');
- GET_PCIADDR_FIELD(input, dev_addr->devid, UINT8_MAX, '.');
- GET_PCIADDR_FIELD(input, dev_addr->function, UINT8_MAX, 0);
+ const char *in = input;
+ unsigned long val;
+ char *end;
+
+ errno = 0;
+ val = strtoul(in, &end, 16);
+ if (errno != 0 || end[0] != ':' || val > UINT16_MAX)
+ return -EINVAL;
+ dev_addr->domain = (uint16_t)val;
+ in = end + 1;
+ in = get_u8_pciaddr_field(in, &dev_addr->bus, ':');
+ if (in == NULL)
+ return -EINVAL;
+ in = get_u8_pciaddr_field(in, &dev_addr->devid, '.');
+ if (in == NULL)
+ return -EINVAL;
+ in = get_u8_pciaddr_field(in, &dev_addr->function, '\0');
+ if (in == NULL)
+ return -EINVAL;
return 0;
}
--
2.1.4
[dpdk-dev] [PATCH v5 14/20] pci: make specialized parsing functions private
Do not expose the minute implementations of PCI parsing.
This leaves only the all-purpose pci_addr_parse, which is simpler to
use.
Signed-off-by: Gaetan Rivet
---
lib/librte_pci/include/rte_pci.h | 29 -
lib/librte_pci/rte_pci.c | 28 ++--
lib/librte_pci/rte_pci_version.map | 2 --
3 files changed, 14 insertions(+), 45 deletions(-)
diff --git a/lib/librte_pci/include/rte_pci.h b/lib/librte_pci/include/rte_pci.h
index 9e79557..ea0897c 100644
--- a/lib/librte_pci/include/rte_pci.h
+++ b/lib/librte_pci/include/rte_pci.h
@@ -151,21 +151,6 @@ TAILQ_HEAD(mapped_pci_res_list, mapped_pci_resource);
int eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr);
/**
- * Utility function to produce a PCI Bus-Device-Function value
- * given a string representation. Assumes that the BDF is provided without
- * a domain prefix (i.e. domain returned is always 0)
- *
- * @param input
- * The input string to be parsed. Should have the format XX:XX.X
- * @param dev_addr
- * The PCI Bus-Device-Function address to be returned.
- * Domain will always be returned as 0
- * @return
- * 0 on success, negative on error.
- */
-int pci_bdf_parse(const char *input, struct rte_pci_addr *dev_addr);
-
-/**
* @deprecated
* Utility function to produce a PCI Bus-Device-Function value
* given a string representation. Assumes that the BDF is provided including
@@ -181,20 +166,6 @@ int pci_bdf_parse(const char *input, struct rte_pci_addr
*dev_addr);
int eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr);
/**
- * Utility function to produce a PCI Bus-Device-Function value
- * given a string representation. Assumes that the BDF is provided including
- * a domain prefix.
- *
- * @param input
- * The input string to be parsed. Should have the format :XX:XX.X
- * @param dev_addr
- * The PCI Bus-Device-Function address to be returned
- * @return
- * 0 on success, negative on error.
- */
-int pci_dbdf_parse(const char *input, struct rte_pci_addr *dev_addr);
-
-/**
* @deprecated
* Utility function to write a pci device name, this device name can later be
* used to retrieve the corresponding rte_pci_addr using eal_parse_pci_*
diff --git a/lib/librte_pci/rte_pci.c b/lib/librte_pci/rte_pci.c
index 7ba472c..1307a18 100644
--- a/lib/librte_pci/rte_pci.c
+++ b/lib/librte_pci/rte_pci.c
@@ -70,13 +70,7 @@ get_u8_pciaddr_field(const char *in, void *_u8, char dlm)
return end + 1;
}
-int
-eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr)
-{
- return pci_bdf_parse(input, dev_addr);
-}
-
-int
+static int
pci_bdf_parse(const char *input, struct rte_pci_addr *dev_addr)
{
const char *in = input;
@@ -94,13 +88,7 @@ pci_bdf_parse(const char *input, struct rte_pci_addr
*dev_addr)
return 0;
}
-int
-eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr)
-{
- return pci_dbdf_parse(input, dev_addr);
-}
-
-int
+static int
pci_dbdf_parse(const char *input, struct rte_pci_addr *dev_addr)
{
const char *in = input;
@@ -125,6 +113,18 @@ pci_dbdf_parse(const char *input, struct rte_pci_addr
*dev_addr)
return 0;
}
+int
+eal_parse_pci_BDF(const char *input, struct rte_pci_addr *dev_addr)
+{
+ return pci_bdf_parse(input, dev_addr);
+}
+
+int
+eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr)
+{
+ return pci_dbdf_parse(input, dev_addr);
+}
+
void
rte_pci_device_name(const struct rte_pci_addr *addr,
char *output, size_t size)
diff --git a/lib/librte_pci/rte_pci_version.map
b/lib/librte_pci/rte_pci_version.map
index 5f50f98..b5c9ec2 100644
--- a/lib/librte_pci/rte_pci_version.map
+++ b/lib/librte_pci/rte_pci_version.map
@@ -5,8 +5,6 @@ DPDK_17.11 {
eal_parse_pci_DomBDF;
pci_addr_cmp;
pci_addr_parse;
- pci_bdf_parse;
- pci_dbdf_parse;
pci_device_name;
pci_map_resource;
pci_unmap_resource;
--
2.1.4
[dpdk-dev] [PATCH v5 13/20] pci: introduce PCI address parsing function
A new single function that is able to parse all currently supported
format:
* Domain-Bus-Device-Function
*Bus-Device-Function
Signed-off-by: Gaetan Rivet
---
lib/librte_pci/include/rte_pci.h | 15 +++
lib/librte_pci/rte_pci.c | 10 ++
lib/librte_pci/rte_pci_version.map | 1 +
3 files changed, 26 insertions(+)
diff --git a/lib/librte_pci/include/rte_pci.h b/lib/librte_pci/include/rte_pci.h
index 38cbdb3..9e79557 100644
--- a/lib/librte_pci/include/rte_pci.h
+++ b/lib/librte_pci/include/rte_pci.h
@@ -256,6 +256,21 @@ int rte_eal_compare_pci_addr(const struct rte_pci_addr
*addr,
int pci_addr_cmp(const struct rte_pci_addr *addr,
const struct rte_pci_addr *addr2);
+
+/**
+ * Utility function to parse a string into a PCI location.
+ *
+ * @param str
+ * The string to parse
+ * @param addr
+ * The reference to the structure where the location
+ * is stored.
+ * @return
+ * 0 on success
+ * <0 otherwise
+ */
+int pci_addr_parse(const char *str, struct rte_pci_addr *addr);
+
/**
* Map a particular resource from a file.
*
diff --git a/lib/librte_pci/rte_pci.c b/lib/librte_pci/rte_pci.c
index fc2fc7f..7ba472c 100644
--- a/lib/librte_pci/rte_pci.c
+++ b/lib/librte_pci/rte_pci.c
@@ -171,6 +171,16 @@ pci_addr_cmp(const struct rte_pci_addr *addr,
return 0;
}
+int
+pci_addr_parse(const char *str, struct rte_pci_addr *addr)
+{
+ if (pci_bdf_parse(str, addr) == 0 ||
+ pci_dbdf_parse(str, addr) == 0)
+ return 0;
+ return -1;
+}
+
+
/* map a particular resource from a file */
void *
pci_map_resource(void *requested_addr, int fd, off_t offset, size_t size,
diff --git a/lib/librte_pci/rte_pci_version.map
b/lib/librte_pci/rte_pci_version.map
index 8d180e3..5f50f98 100644
--- a/lib/librte_pci/rte_pci_version.map
+++ b/lib/librte_pci/rte_pci_version.map
@@ -4,6 +4,7 @@ DPDK_17.11 {
eal_parse_pci_BDF;
eal_parse_pci_DomBDF;
pci_addr_cmp;
+ pci_addr_parse;
pci_bdf_parse;
pci_dbdf_parse;
pci_device_name;
--
2.1.4
[dpdk-dev] [PATCH v5 12/20] pci: deprecate misnamed functions
Rename misnamed functions and describe the change in a deprecation
notice.
Signed-off-by: Gaetan Rivet
---
doc/guides/rel_notes/deprecation.rst | 10 ++
lib/librte_pci/include/rte_pci.h | 63
lib/librte_pci/rte_pci.c | 26 +++
lib/librte_pci/rte_pci_version.map | 4 +++
4 files changed, 103 insertions(+)
diff --git a/doc/guides/rel_notes/deprecation.rst
b/doc/guides/rel_notes/deprecation.rst
index 4e4d97b..ef2264f 100644
--- a/doc/guides/rel_notes/deprecation.rst
+++ b/doc/guides/rel_notes/deprecation.rst
@@ -98,3 +98,13 @@ Deprecation Notices
The non-"do-sig" versions of the hash tables will be removed
(including the ``signature_offset`` parameter)
and the "do-sig" versions renamed accordingly.
+
+* pci: Several exposed functions are misnamed.
+ The following functions are deprecated starting from v17.11 and are replaced:
+
+ - ``eal_parse_pci_BDF`` replaced by ``pci_parse_BDF``
+ - ``eal_parse_pci_DomBDF`` replaced by ``pci_parse_DomBDF``
+ - ``rte_eal_compare_pci_addr`` replaced by ``pci_addr_cmp``
+ - ``rte_pci_device_name`` replaced by ``pci_device_name``
+
+ The functions are only renamed. Their behavior is not affected.
diff --git a/lib/librte_pci/include/rte_pci.h b/lib/librte_pci/include/rte_pci.h
index fe4d411..38cbdb3 100644
--- a/lib/librte_pci/include/rte_pci.h
+++ b/lib/librte_pci/include/rte_pci.h
@@ -135,6 +135,7 @@ struct mapped_pci_resource {
TAILQ_HEAD(mapped_pci_res_list, mapped_pci_resource);
/**
+ * @deprecated
* Utility function to produce a PCI Bus-Device-Function value
* given a string representation. Assumes that the BDF is provided without
* a domain prefix (i.e. domain returned is always 0)
@@ -151,6 +152,22 @@ int eal_parse_pci_BDF(const char *input, struct
rte_pci_addr *dev_addr);
/**
* Utility function to produce a PCI Bus-Device-Function value
+ * given a string representation. Assumes that the BDF is provided without
+ * a domain prefix (i.e. domain returned is always 0)
+ *
+ * @param input
+ * The input string to be parsed. Should have the format XX:XX.X
+ * @param dev_addr
+ * The PCI Bus-Device-Function address to be returned.
+ * Domain will always be returned as 0
+ * @return
+ * 0 on success, negative on error.
+ */
+int pci_bdf_parse(const char *input, struct rte_pci_addr *dev_addr);
+
+/**
+ * @deprecated
+ * Utility function to produce a PCI Bus-Device-Function value
* given a string representation. Assumes that the BDF is provided including
* a domain prefix.
*
@@ -164,6 +181,21 @@ int eal_parse_pci_BDF(const char *input, struct
rte_pci_addr *dev_addr);
int eal_parse_pci_DomBDF(const char *input, struct rte_pci_addr *dev_addr);
/**
+ * Utility function to produce a PCI Bus-Device-Function value
+ * given a string representation. Assumes that the BDF is provided including
+ * a domain prefix.
+ *
+ * @param input
+ * The input string to be parsed. Should have the format :XX:XX.X
+ * @param dev_addr
+ * The PCI Bus-Device-Function address to be returned
+ * @return
+ * 0 on success, negative on error.
+ */
+int pci_dbdf_parse(const char *input, struct rte_pci_addr *dev_addr);
+
+/**
+ * @deprecated
* Utility function to write a pci device name, this device name can later be
* used to retrieve the corresponding rte_pci_addr using eal_parse_pci_*
* BDF helpers.
@@ -179,6 +211,22 @@ void rte_pci_device_name(const struct rte_pci_addr *addr,
char *output, size_t size);
/**
+ * Utility function to write a pci device name, this device name can later be
+ * used to retrieve the corresponding rte_pci_addr using eal_parse_pci_*
+ * BDF helpers.
+ *
+ * @param addr
+ * The PCI Bus-Device-Function address
+ * @param output
+ * The output buffer string
+ * @param size
+ * The output buffer size
+ */
+void pci_device_name(const struct rte_pci_addr *addr,
+char *output, size_t size);
+
+/**
+ * @deprecated
* Utility function to compare two PCI device addresses.
*
* @param addr
@@ -194,6 +242,21 @@ int rte_eal_compare_pci_addr(const struct rte_pci_addr
*addr,
const struct rte_pci_addr *addr2);
/**
+ * Utility function to compare two PCI device addresses.
+ *
+ * @param addr
+ * The PCI Bus-Device-Function address to compare
+ * @param addr2
+ * The PCI Bus-Device-Function address to compare
+ * @return
+ * 0 on equal PCI address.
+ * Positive on addr is greater than addr2.
+ * Negative on addr is less than addr2, or error.
+ */
+int pci_addr_cmp(const struct rte_pci_addr *addr,
+const struct rte_pci_addr *addr2);
+
+/**
* Map a particular resource from a file.
*
* @param requested_addr
diff --git a/lib/librte_pci/rte_pci.c b/lib/librte_pci/rte_pci.c
index cbb5359..fc2fc7f 100644
--- a/lib/librte_pci/rte_pci.c
+++ b/lib/librte_pci/rte_pci.c
@@ -73,6 +73,12 @@ get_u8_pciaddr_field(const char *in,
[dpdk-dev] [PATCH v5 19/20] doc: add notes on EAL PCI API update
Add a section related to EAL API changes to 17.11 release notes. Signed-off-by: Gaetan Rivet Acked-by: John McNamara --- doc/guides/rel_notes/release_17_11.rst | 28 1 file changed, 28 insertions(+) diff --git a/doc/guides/rel_notes/release_17_11.rst b/doc/guides/rel_notes/release_17_11.rst index 63b9869..ed3d6d8 100644 --- a/doc/guides/rel_notes/release_17_11.rst +++ b/doc/guides/rel_notes/release_17_11.rst @@ -232,6 +232,34 @@ API Changes By this way PMDs can return an error value in case of failure at stats getting process time. +* **PCI bus API moved outside of the EAL** + + The PCI bus previously implemented within the EAL has been moved. + A first part has been added as an RTE library providing PCI helpers to + parse device locations or other such utilities. + A second part consisting in the actual bus driver has been moved to its + proper subdirectory, without changing its functionalities. + + As such, several PCI-related functions are not proposed by the EAL anymore: + + * rte_pci_detach + * rte_pci_dump + * rte_pci_ioport_map + * rte_pci_ioport_read + * rte_pci_ioport_unmap + * rte_pci_ioport_write + * rte_pci_map_device + * rte_pci_probe + * rte_pci_probe_one + * rte_pci_read_config + * rte_pci_register + * rte_pci_scan + * rte_pci_unmap_device + * rte_pci_unregister + * rte_pci_write_config + + These functions are made available either as part of ``librte_pci`` or + ``librte_bus_pci``. ABI Changes --- -- 2.1.4
[dpdk-dev] [PATCH v5 18/20] bus/pci: do not expose IOVA mode getter
Signed-off-by: Gaetan Rivet
---
drivers/bus/pci/private.h | 10 ++
lib/librte_eal/bsdapp/eal/rte_eal_version.map | 1 -
lib/librte_eal/linuxapp/eal/rte_eal_version.map | 1 -
3 files changed, 10 insertions(+), 2 deletions(-)
diff --git a/drivers/bus/pci/private.h b/drivers/bus/pci/private.h
index b91217b..2283f09 100644
--- a/drivers/bus/pci/private.h
+++ b/drivers/bus/pci/private.h
@@ -235,4 +235,14 @@ int
rte_pci_match(const struct rte_pci_driver *pci_drv,
const struct rte_pci_device *pci_dev);
+/**
+ * Get iommu class of PCI devices on the bus.
+ * And return their preferred iova mapping mode.
+ *
+ * @return
+ * - enum rte_iova_mode.
+ */
+enum rte_iova_mode
+rte_pci_get_iommu_class(void);
+
#endif /* _PCI_PRIVATE_H_ */
diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
index 1c18ca5..1d879a1 100644
--- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
@@ -228,6 +228,5 @@ DPDK_17.11 {
rte_eal_iova_mode;
rte_eal_mbuf_default_mempool_ops;
rte_lcore_has_role;
- rte_pci_get_iommu_class;
} DPDK_17.08;
diff --git a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
index 827bcd4..1334433 100644
--- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
@@ -232,6 +232,5 @@ DPDK_17.11 {
rte_eal_iova_mode;
rte_eal_mbuf_default_mempool_ops;
rte_lcore_has_role;
- rte_pci_get_iommu_class;
} DPDK_17.08;
--
2.1.4
[dpdk-dev] [PATCH v5 20/20] bus: rename scan policy as probe policy
This bus configuration item is misnamed, as it actually refers to the
probing process.
Signed-off-by: Gaetan Rivet
---
drivers/bus/pci/pci_common.c | 2 +-
lib/librte_eal/common/eal_common_devargs.c | 6 +++---
lib/librte_eal/common/include/rte_bus.h| 12 ++--
3 files changed, 10 insertions(+), 10 deletions(-)
diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c
index d7a1c05..cc23a39 100644
--- a/drivers/bus/pci/pci_common.c
+++ b/drivers/bus/pci/pci_common.c
@@ -373,7 +373,7 @@ rte_pci_probe(void)
int probe_all = 0;
int ret = 0;
- if (rte_pci_bus.bus.conf.scan_mode != RTE_BUS_SCAN_WHITELIST)
+ if (rte_pci_bus.bus.conf.probe_mode != RTE_BUS_PROBE_WHITELIST)
probe_all = 1;
FOREACH_DEVICE_ON_PCIBUS(dev) {
diff --git a/lib/librte_eal/common/eal_common_devargs.c
b/lib/librte_eal/common/eal_common_devargs.c
index 6ac88d6..f5ef913 100644
--- a/lib/librte_eal/common/eal_common_devargs.c
+++ b/lib/librte_eal/common/eal_common_devargs.c
@@ -170,11 +170,11 @@ rte_eal_devargs_add(enum rte_devtype devtype, const char
*devargs_str)
bus = devargs->bus;
if (devargs->type == RTE_DEVTYPE_BLACKLISTED_PCI)
devargs->policy = RTE_DEV_BLACKLISTED;
- if (bus->conf.scan_mode == RTE_BUS_SCAN_UNDEFINED) {
+ if (bus->conf.probe_mode == RTE_BUS_PROBE_UNDEFINED) {
if (devargs->policy == RTE_DEV_WHITELISTED)
- bus->conf.scan_mode = RTE_BUS_SCAN_WHITELIST;
+ bus->conf.probe_mode = RTE_BUS_PROBE_WHITELIST;
else if (devargs->policy == RTE_DEV_BLACKLISTED)
- bus->conf.scan_mode = RTE_BUS_SCAN_BLACKLIST;
+ bus->conf.probe_mode = RTE_BUS_PROBE_BLACKLIST;
}
TAILQ_INSERT_TAIL(&devargs_list, devargs, next);
return 0;
diff --git a/lib/librte_eal/common/include/rte_bus.h
b/lib/librte_eal/common/include/rte_bus.h
index 6fb0834..331d954 100644
--- a/lib/librte_eal/common/include/rte_bus.h
+++ b/lib/librte_eal/common/include/rte_bus.h
@@ -168,19 +168,19 @@ typedef int (*rte_bus_unplug_t)(struct rte_device *dev);
typedef int (*rte_bus_parse_t)(const char *name, void *addr);
/**
- * Bus scan policies
+ * Bus probe policies
*/
-enum rte_bus_scan_mode {
- RTE_BUS_SCAN_UNDEFINED,
- RTE_BUS_SCAN_WHITELIST,
- RTE_BUS_SCAN_BLACKLIST,
+enum rte_bus_probe_mode {
+ RTE_BUS_PROBE_UNDEFINED,
+ RTE_BUS_PROBE_WHITELIST,
+ RTE_BUS_PROBE_BLACKLIST,
};
/**
* A structure used to configure bus operations.
*/
struct rte_bus_conf {
- enum rte_bus_scan_mode scan_mode; /**< Scan policy. */
+ enum rte_bus_probe_mode probe_mode; /**< Probe policy. */
};
--
2.1.4
[dpdk-dev] [PATCH v5 17/20] bus/pci: do not expose PCI match function
This function is private to the PCI bus.
Signed-off-by: Gaetan Rivet
---
drivers/bus/pci/include/rte_bus_pci.h | 16
drivers/bus/pci/private.h | 15 +++
lib/librte_eal/bsdapp/eal/rte_eal_version.map | 1 -
lib/librte_eal/linuxapp/eal/rte_eal_version.map | 1 -
4 files changed, 15 insertions(+), 18 deletions(-)
diff --git a/drivers/bus/pci/include/rte_bus_pci.h
b/drivers/bus/pci/include/rte_bus_pci.h
index d188353..f0a3b37 100644
--- a/drivers/bus/pci/include/rte_bus_pci.h
+++ b/drivers/bus/pci/include/rte_bus_pci.h
@@ -170,22 +170,6 @@ struct rte_pci_bus {
/** Device driver supports IOVA as VA */
#define RTE_PCI_DRV_IOVA_AS_VA 0X0040
-/*
- * Match the PCI Driver and Device using the ID Table
- *
- * @param pci_drv
- * PCI driver from which ID table would be extracted
- * @param pci_dev
- * PCI device to match against the driver
- * @return
- * 1 for successful match
- * 0 for unsuccessful match
- */
-int
-rte_pci_match(const struct rte_pci_driver *pci_drv,
- const struct rte_pci_device *pci_dev);
-
-
/**
* Get iommu class of PCI devices on the bus.
* And return their preferred iova mapping mode.
diff --git a/drivers/bus/pci/private.h b/drivers/bus/pci/private.h
index 455c81d..b91217b 100644
--- a/drivers/bus/pci/private.h
+++ b/drivers/bus/pci/private.h
@@ -220,4 +220,19 @@ void pci_uio_free_resource(struct rte_pci_device *dev,
int pci_uio_map_resource_by_index(struct rte_pci_device *dev, int res_idx,
struct mapped_pci_resource *uio_res, int map_idx);
+/*
+ * Match the PCI Driver and Device using the ID Table
+ *
+ * @param pci_drv
+ * PCI driver from which ID table would be extracted
+ * @param pci_dev
+ * PCI device to match against the driver
+ * @return
+ * 1 for successful match
+ * 0 for unsuccessful match
+ */
+int
+rte_pci_match(const struct rte_pci_driver *pci_drv,
+ const struct rte_pci_device *pci_dev);
+
#endif /* _PCI_PRIVATE_H_ */
diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
index 97b3918..1c18ca5 100644
--- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
@@ -229,6 +229,5 @@ DPDK_17.11 {
rte_eal_mbuf_default_mempool_ops;
rte_lcore_has_role;
rte_pci_get_iommu_class;
- rte_pci_match;
} DPDK_17.08;
diff --git a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
index a8ea4ea..827bcd4 100644
--- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
@@ -233,6 +233,5 @@ DPDK_17.11 {
rte_eal_mbuf_default_mempool_ops;
rte_lcore_has_role;
rte_pci_get_iommu_class;
- rte_pci_match;
} DPDK_17.08;
--
2.1.4
[dpdk-dev] [PATCH v5 16/20] bus/pci: do not expose private functions
make the functions
+ rte_pci_detach
+ rte_pci_probe
+ rte_pci_probe_one
+ rte_pci_scan
private as there is no point in using them outside of the rte_bus
framework.
Signed-off-by: Gaetan Rivet
---
drivers/bus/pci/include/rte_bus_pci.h | 49 -
drivers/bus/pci/private.h | 49 +
drivers/bus/pci/rte_bus_pci_version.map | 4 ---
3 files changed, 49 insertions(+), 53 deletions(-)
diff --git a/drivers/bus/pci/include/rte_bus_pci.h
b/drivers/bus/pci/include/rte_bus_pci.h
index e6a7998..d188353 100644
--- a/drivers/bus/pci/include/rte_bus_pci.h
+++ b/drivers/bus/pci/include/rte_bus_pci.h
@@ -170,25 +170,6 @@ struct rte_pci_bus {
/** Device driver supports IOVA as VA */
#define RTE_PCI_DRV_IOVA_AS_VA 0X0040
-/**
- * Scan the content of the PCI bus, and the devices in the devices
- * list
- *
- * @return
- * 0 on success, negative on error
- */
-int rte_pci_scan(void);
-
-/**
- * Probe the PCI bus
- *
- * @return
- * - 0 on success.
- * - !0 on error.
- */
-int
-rte_pci_probe(void);
-
/*
* Match the PCI Driver and Device using the ID Table
*
@@ -242,36 +223,6 @@ int rte_pci_map_device(struct rte_pci_device *dev);
void rte_pci_unmap_device(struct rte_pci_device *dev);
/**
- * Probe the single PCI device.
- *
- * Scan the content of the PCI bus, and find the pci device specified by pci
- * address, then call the probe() function for registered driver that has a
- * matching entry in its id_table for discovered device.
- *
- * @param addr
- * The PCI Bus-Device-Function address to probe.
- * @return
- * - 0 on success.
- * - Negative on error.
- */
-int rte_pci_probe_one(const struct rte_pci_addr *addr);
-
-/**
- * Close the single PCI device.
- *
- * Scan the content of the PCI bus, and find the pci device specified by pci
- * address, then call the remove() function for registered driver that has a
- * matching entry in its id_table for discovered device.
- *
- * @param addr
- * The PCI Bus-Device-Function address to close.
- * @return
- * - 0 on success.
- * - Negative on error.
- */
-int rte_pci_detach(const struct rte_pci_addr *addr);
-
-/**
* Dump the content of the PCI bus.
*
* @param f
diff --git a/drivers/bus/pci/private.h b/drivers/bus/pci/private.h
index fdc2c81..455c81d 100644
--- a/drivers/bus/pci/private.h
+++ b/drivers/bus/pci/private.h
@@ -43,6 +43,55 @@ struct rte_pci_driver;
struct rte_pci_device;
/**
+ * Probe the PCI bus
+ *
+ * @return
+ * - 0 on success.
+ * - !0 on error.
+ */
+int
+rte_pci_probe(void);
+
+/**
+ * Scan the content of the PCI bus, and the devices in the devices
+ * list
+ *
+ * @return
+ * 0 on success, negative on error
+ */
+int rte_pci_scan(void);
+
+/**
+ * Probe the single PCI device.
+ *
+ * Scan the content of the PCI bus, and find the pci device specified by pci
+ * address, then call the probe() function for registered driver that has a
+ * matching entry in its id_table for discovered device.
+ *
+ * @param addr
+ * The PCI Bus-Device-Function address to probe.
+ * @return
+ * - 0 on success.
+ * - Negative on error.
+ */
+int rte_pci_probe_one(const struct rte_pci_addr *addr);
+
+/**
+ * Close the single PCI device.
+ *
+ * Scan the content of the PCI bus, and find the pci device specified by pci
+ * address, then call the remove() function for registered driver that has a
+ * matching entry in its id_table for discovered device.
+ *
+ * @param addr
+ * The PCI Bus-Device-Function address to close.
+ * @return
+ * - 0 on success.
+ * - Negative on error.
+ */
+int rte_pci_detach(const struct rte_pci_addr *addr);
+
+/**
* Find the name of a PCI device.
*/
void
diff --git a/drivers/bus/pci/rte_bus_pci_version.map
b/drivers/bus/pci/rte_bus_pci_version.map
index c1973f6..ee67033 100644
--- a/drivers/bus/pci/rte_bus_pci_version.map
+++ b/drivers/bus/pci/rte_bus_pci_version.map
@@ -1,18 +1,14 @@
DPDK_17.11 {
global:
- rte_pci_detach;
rte_pci_dump;
rte_pci_ioport_map;
rte_pci_ioport_read;
rte_pci_ioport_unmap;
rte_pci_ioport_write;
rte_pci_map_device;
- rte_pci_probe;
- rte_pci_probe_one;
rte_pci_read_config;
rte_pci_register;
- rte_pci_scan;
rte_pci_unmap_device;
rte_pci_unregister;
rte_pci_write_config;
--
2.1.4
Re: [dpdk-dev] [PATCH V3 3/5] ethdev: add new api for traffic metering and policing
On 10/6/2017 8:15 PM, Cristian Dumitrescu wrote:
Overall the patch looks ok. Few small comments.
diff --git a/lib/librte_ether/rte_mtr.h b/lib/librte_ether/rte_mtr.h
new file mode 100644
index 000..f0f897f
--- /dev/null
+++ b/lib/librte_ether/rte_mtr.h
@@ -0,0 +1,723 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2017 Intel Corporation.
+ * Copyright(c) 2017 NXP.
can you make it "Copyright 2017 NXP"
+ * Copyright(c) 2017 Cavium.
+ * All rights reserved.
+ *
...
+/**
+ * Statistics counter type
+ */
+enum rte_mtr_stats_type {
+ /** Number of packets passed as green by the policer. */
+ RTE_MTR_STATS_N_PKTS_GREEN = 1 << 0,
+
+ /** Number of bytes passed as green by the policer. */
+ RTE_MTR_STATS_N_BYTES_GREEN = 1 << 1,
+
+ /** Number of packets passed as yellow by the policer. */
+ RTE_MTR_STATS_N_PKTS_YELLOW = 1 << 2,
+
+ /** Number of bytes passed as yellow by the policer. */
+ RTE_MTR_STATS_N_BYTES_YELLOW = 1 << 3,
+
+ /** Number of packets passed as red by the policer. */
+ RTE_MTR_STATS_N_PKTS_RED = 1 << 4,
+
+ /** Number of bytes passed as red by the policer. */
+ RTE_MTR_STATS_N_BYTES_RED = 1 << 5,
+
+ /** Number of packets dropped by the policer. */
+ RTE_MTR_STATS_N_PKTS_DROPPED = 1 << 6,
+
+ /** Number of bytes dropped by the policer. */
+ RTE_MTR_STATS_N_BYTES_DROPPED = 1 << 7,
Trivial one.
do you want to group the pkts and bytes in separate group respectively?
+};
+
+/**
+ * Statistics counters
+ */
+struct rte_mtr_stats {
+ /** Number of packets passed by the policer (per color). */
+ uint64_t n_pkts[RTE_MTR_COLORS];
+
+ /** Number of bytes passed by the policer (per color). */
+ uint64_t n_bytes[RTE_MTR_COLORS];
+
+ /** Number of packets dropped by the policer. */
+ uint64_t n_pkts_dropped;
+
+ /** Number of bytes passed by the policer. */
+ uint64_t n_bytes_dropped;
+};
+
+/**
+ * Traffic metering algorithms
+ */
+enum rte_mtr_algorithm {
+ /** Single Rate Three Color Marker (srTCM) - IETF RFC 2697. */
+ RTE_MTR_SRTCM_RFC2697,
+
+ /** Two Rate Three Color Marker (trTCM) - IETF RFC 2698. */
+ RTE_MTR_TRTCM_RFC2698,
+
+ /** Two Rate Three Color Marker (trTCM) - IETF RFC 4115. */
+ RTE_MTR_TRTCM_RFC4115,
I will still prefer if you still add "PASS_THROUGH" as explicit algo.
+};
+
+/**
+ * Meter profile
+ */
Re: [dpdk-dev] [PATCH v7 5/6] igb_uio: use kernel functions for masking MSI-X
Hi,
This commit renders igb_uio unusable on Fedora 20, kernel version
3.19.8-100.fc20.x86_64.
During the build (make -j 20), a warning is issued for igb_uio regarding a
missing symbol (pci_msi_unmask):
WARNING: "pci_msi_unmask_irq"
[/home//x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/igb_uio/igb_uio.ko]
undefined!
WARNING: "pci_msi_mask_irq"
[/home//x86_64-native-linuxapp-gcc/build/lib/librte_eal/linuxapp/igb_uio/igb_uio.ko]
undefined!
Subsequently, the module is may not be loaded on account of same.
(from dmesg):
[673425.712110] igb_uio: Unknown symbol pci_msi_unmask_irq (err 0)
[673425.712124] igb_uio: Unknown symbol pci_msi_mask_irq (err 0)
Thanks,
Mark
>-Original Message-
>From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Markus Theil
>Sent: Tuesday, September 5, 2017 1:04 PM
>To: dev@dpdk.org
>Cc: Yigit, Ferruh ; step...@networkplumber.org; Markus
>Theil
>Subject: [dpdk-dev] [PATCH v7 5/6] igb_uio: use kernel functions for masking
>MSI-X
>
>This patch removes the custom MSI-X mask/unmask code and
>uses already existing kernel functions.
>
>Signed-off-by: Markus Theil
>---
> lib/librte_eal/linuxapp/igb_uio/compat.h | 26 +---
> lib/librte_eal/linuxapp/igb_uio/igb_uio.c | 51 --
>-
> 2 files changed, 28 insertions(+), 49 deletions(-)
>
>diff --git a/lib/librte_eal/linuxapp/igb_uio/compat.h
>b/lib/librte_eal/linuxapp/igb_uio/compat.h
>index 3825933..67a7ab3 100644
>--- a/lib/librte_eal/linuxapp/igb_uio/compat.h
>+++ b/lib/librte_eal/linuxapp/igb_uio/compat.h
>@@ -15,24 +15,6 @@
> #define HAVE_PTE_MASK_PAGE_IOMAP
> #endif
>
>-#ifndef PCI_MSIX_ENTRY_SIZE
>-#define PCI_MSIX_ENTRY_SIZE 16
>-#define PCI_MSIX_ENTRY_LOWER_ADDR 0
>-#define PCI_MSIX_ENTRY_UPPER_ADDR 4
>-#define PCI_MSIX_ENTRY_DATA8
>-#define PCI_MSIX_ENTRY_VECTOR_CTRL 12
>-#define PCI_MSIX_ENTRY_CTRL_MASKBIT 1
>-#endif
>-
>-/*
>- * for kernels < 2.6.38 and backported patch that moves MSI-X entry
>definition
>- * to pci_regs.h Those kernels has PCI_MSIX_ENTRY_SIZE defined but not
>- * PCI_MSIX_ENTRY_CTRL_MASKBIT
>- */
>-#ifndef PCI_MSIX_ENTRY_CTRL_MASKBIT
>-#define PCI_MSIX_ENTRY_CTRL_MASKBIT1
>-#endif
>-
> #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 34) && \
> (!(defined(RHEL_RELEASE_CODE) && \
>RHEL_RELEASE_CODE >= RHEL_RELEASE_VERSION(5, 9)))
>@@ -127,3 +109,11 @@ static bool pci_check_and_mask_intx(struct pci_dev *pdev)
> #if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 8, 0)
> #define HAVE_ALLOC_IRQ_VECTORS 1
> #endif
>+
>+#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 19, 0)
>+#define HAVE_PCI_MSI_MASK_IRQ 1
>+#endif
>+
>+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 37)
>+#define HAVE_IRQ_DATA 1
>+#endif
>diff --git a/lib/librte_eal/linuxapp/igb_uio/igb_uio.c
>b/lib/librte_eal/linuxapp/igb_uio/igb_uio.c
>index c570eed..e4ef817 100644
>--- a/lib/librte_eal/linuxapp/igb_uio/igb_uio.c
>+++ b/lib/librte_eal/linuxapp/igb_uio/igb_uio.c
>@@ -91,27 +91,6 @@ static struct attribute *dev_attrs[] = {
> static const struct attribute_group dev_attr_grp = {
> .attrs = dev_attrs,
> };
>-/*
>- * It masks the msix on/off of generating MSI-X messages.
>- */
>-static void
>-igbuio_msix_mask_irq(struct msi_desc *desc, int32_t state)
>-{
>- u32 mask_bits = desc->masked;
>- unsigned offset = desc->msi_attrib.entry_nr * PCI_MSIX_ENTRY_SIZE +
>- PCI_MSIX_ENTRY_VECTOR_CTRL;
>-
>- if (state != 0)
>- mask_bits &= ~PCI_MSIX_ENTRY_CTRL_MASKBIT;
>- else
>- mask_bits |= PCI_MSIX_ENTRY_CTRL_MASKBIT;
>-
>- if (mask_bits != desc->masked) {
>- writel(mask_bits, desc->mask_base + offset);
>- readl(desc->mask_base);
>- desc->masked = mask_bits;
>- }
>-}
>
> /**
> * This is the irqcontrol callback to be registered to uio_info.
>@@ -132,21 +111,31 @@ igbuio_pci_irqcontrol(struct uio_info *info, s32
>irq_state)
> struct rte_uio_pci_dev *udev = info->priv;
> struct pci_dev *pdev = udev->pdev;
>
>- pci_cfg_access_lock(pdev);
>- if (udev->mode == RTE_INTR_MODE_LEGACY)
>- pci_intx(pdev, !!irq_state);
>+#ifdef HAVE_IRQ_DATA
>+ struct irq_data *irq = irq_get_irq_data(udev->info.irq);
>+#else
>+ unsigned int irq = udev->info.irq;
>+#endif
>
>- else if (udev->mode == RTE_INTR_MODE_MSIX) {
>- struct msi_desc *desc;
>+ pci_cfg_access_lock(pdev);
>
>-#if (LINUX_VERSION_CODE < KERNEL_VERSION(4, 3, 0))
>- list_for_each_entry(desc, &pdev->msi_list, list)
>- igbuio_msix_mask_irq(desc, irq_state);
>+ if (udev->mode == RTE_INTR_MODE_MSIX) {
>+#ifdef HAVE_PCI_MSI_MASK_IRQ
>+ if (irq_state == 1)
>+ pci_msi_unmask_irq(irq);
>+ else
>+ pci_msi_mask_irq(irq);
> #else
>- list_for_each_entry(desc, &pdev->dev.ms
Re: [dpdk-dev] [PATCH V3 3/5] ethdev: add new api for traffic metering and policing
On 10/12/2017 4:18 PM, Hemant Agrawal wrote:
On 10/6/2017 8:15 PM, Cristian Dumitrescu wrote:
Overall the patch looks ok. Few small comments.
diff --git a/lib/librte_ether/rte_mtr.h b/lib/librte_ether/rte_mtr.h
new file mode 100644
index 000..f0f897f
--- /dev/null
+++ b/lib/librte_ether/rte_mtr.h
@@ -0,0 +1,723 @@
+/*-
+ * BSD LICENSE
+ *
+ * Copyright(c) 2017 Intel Corporation.
+ * Copyright(c) 2017 NXP.
can you make it "Copyright 2017 NXP"
+ * Copyright(c) 2017 Cavium.
+ * All rights reserved.
+ *
...
+/**
+ * Statistics counter type
+ */
+enum rte_mtr_stats_type {
+/** Number of packets passed as green by the policer. */
+RTE_MTR_STATS_N_PKTS_GREEN = 1 << 0,
+
+/** Number of bytes passed as green by the policer. */
+RTE_MTR_STATS_N_BYTES_GREEN = 1 << 1,
+
+/** Number of packets passed as yellow by the policer. */
+RTE_MTR_STATS_N_PKTS_YELLOW = 1 << 2,
+
+/** Number of bytes passed as yellow by the policer. */
+RTE_MTR_STATS_N_BYTES_YELLOW = 1 << 3,
+
+/** Number of packets passed as red by the policer. */
+RTE_MTR_STATS_N_PKTS_RED = 1 << 4,
+
+/** Number of bytes passed as red by the policer. */
+RTE_MTR_STATS_N_BYTES_RED = 1 << 5,
+
+/** Number of packets dropped by the policer. */
+RTE_MTR_STATS_N_PKTS_DROPPED = 1 << 6,
+
+/** Number of bytes dropped by the policer. */
+RTE_MTR_STATS_N_BYTES_DROPPED = 1 << 7,
Trivial one.
do you want to group the pkts and bytes in separate group respectively?
+};
+
+/**
+ * Statistics counters
+ */
+struct rte_mtr_stats {
+/** Number of packets passed by the policer (per color). */
+uint64_t n_pkts[RTE_MTR_COLORS];
+
+/** Number of bytes passed by the policer (per color). */
+uint64_t n_bytes[RTE_MTR_COLORS];
+
+/** Number of packets dropped by the policer. */
+uint64_t n_pkts_dropped;
+
+/** Number of bytes passed by the policer. */
+uint64_t n_bytes_dropped;
+};
+
+/**
+ * Traffic metering algorithms
+ */
+enum rte_mtr_algorithm {
+/** Single Rate Three Color Marker (srTCM) - IETF RFC 2697. */
+RTE_MTR_SRTCM_RFC2697,
+
+/** Two Rate Three Color Marker (trTCM) - IETF RFC 2698. */
+RTE_MTR_TRTCM_RFC2698,
+
+/** Two Rate Three Color Marker (trTCM) - IETF RFC 4115. */
+RTE_MTR_TRTCM_RFC4115,
I will still prefer if you still add "PASS_THROUGH" as explicit algo.
+};
+
+/**
+ * Meter profile
+ */
Reviewed-by: Hemant Agrawal
Re: [dpdk-dev] [PATCH v5 18/20] bus/pci: do not expose IOVA mode getter
On Thu, Oct 12, 2017 at 12:45:45PM +0200, Gaetan Rivet wrote:
> Signed-off-by: Gaetan Rivet
> ---
> drivers/bus/pci/private.h | 10 ++
> lib/librte_eal/bsdapp/eal/rte_eal_version.map | 1 -
> lib/librte_eal/linuxapp/eal/rte_eal_version.map | 1 -
> 3 files changed, 10 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/bus/pci/private.h b/drivers/bus/pci/private.h
> index b91217b..2283f09 100644
> --- a/drivers/bus/pci/private.h
> +++ b/drivers/bus/pci/private.h
> @@ -235,4 +235,14 @@ int
> rte_pci_match(const struct rte_pci_driver *pci_drv,
> const struct rte_pci_device *pci_dev);
>
> +/**
> + * Get iommu class of PCI devices on the bus.
> + * And return their preferred iova mapping mode.
> + *
> + * @return
> + * - enum rte_iova_mode.
> + */
> +enum rte_iova_mode
> +rte_pci_get_iommu_class(void);
> +
Made a mistake here while rebasing.
The function declaration should be removed from
drivers/bus/pci/include/rte_bus_pci.h
as well.
> #endif /* _PCI_PRIVATE_H_ */
> diff --git a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
> b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
> index 1c18ca5..1d879a1 100644
> --- a/lib/librte_eal/bsdapp/eal/rte_eal_version.map
> +++ b/lib/librte_eal/bsdapp/eal/rte_eal_version.map
> @@ -228,6 +228,5 @@ DPDK_17.11 {
> rte_eal_iova_mode;
> rte_eal_mbuf_default_mempool_ops;
> rte_lcore_has_role;
> - rte_pci_get_iommu_class;
>
> } DPDK_17.08;
> diff --git a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
> b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
> index 827bcd4..1334433 100644
> --- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
> +++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
> @@ -232,6 +232,5 @@ DPDK_17.11 {
> rte_eal_iova_mode;
> rte_eal_mbuf_default_mempool_ops;
> rte_lcore_has_role;
> - rte_pci_get_iommu_class;
>
> } DPDK_17.08;
> --
> 2.1.4
>
--
Gaëtan Rivet
6WIND
Re: [dpdk-dev] [PATCH v5 20/20] bus: rename scan policy as probe policy
This patch is not part of this series, sorry.
I can send a v6 if necessary, but here simply ignore this patch.
On Thu, Oct 12, 2017 at 12:45:47PM +0200, Gaetan Rivet wrote:
> This bus configuration item is misnamed, as it actually refers to the
> probing process.
>
> Signed-off-by: Gaetan Rivet
> ---
> drivers/bus/pci/pci_common.c | 2 +-
> lib/librte_eal/common/eal_common_devargs.c | 6 +++---
> lib/librte_eal/common/include/rte_bus.h| 12 ++--
> 3 files changed, 10 insertions(+), 10 deletions(-)
>
> diff --git a/drivers/bus/pci/pci_common.c b/drivers/bus/pci/pci_common.c
> index d7a1c05..cc23a39 100644
> --- a/drivers/bus/pci/pci_common.c
> +++ b/drivers/bus/pci/pci_common.c
> @@ -373,7 +373,7 @@ rte_pci_probe(void)
> int probe_all = 0;
> int ret = 0;
>
> - if (rte_pci_bus.bus.conf.scan_mode != RTE_BUS_SCAN_WHITELIST)
> + if (rte_pci_bus.bus.conf.probe_mode != RTE_BUS_PROBE_WHITELIST)
> probe_all = 1;
>
> FOREACH_DEVICE_ON_PCIBUS(dev) {
> diff --git a/lib/librte_eal/common/eal_common_devargs.c
> b/lib/librte_eal/common/eal_common_devargs.c
> index 6ac88d6..f5ef913 100644
> --- a/lib/librte_eal/common/eal_common_devargs.c
> +++ b/lib/librte_eal/common/eal_common_devargs.c
> @@ -170,11 +170,11 @@ rte_eal_devargs_add(enum rte_devtype devtype, const
> char *devargs_str)
> bus = devargs->bus;
> if (devargs->type == RTE_DEVTYPE_BLACKLISTED_PCI)
> devargs->policy = RTE_DEV_BLACKLISTED;
> - if (bus->conf.scan_mode == RTE_BUS_SCAN_UNDEFINED) {
> + if (bus->conf.probe_mode == RTE_BUS_PROBE_UNDEFINED) {
> if (devargs->policy == RTE_DEV_WHITELISTED)
> - bus->conf.scan_mode = RTE_BUS_SCAN_WHITELIST;
> + bus->conf.probe_mode = RTE_BUS_PROBE_WHITELIST;
> else if (devargs->policy == RTE_DEV_BLACKLISTED)
> - bus->conf.scan_mode = RTE_BUS_SCAN_BLACKLIST;
> + bus->conf.probe_mode = RTE_BUS_PROBE_BLACKLIST;
> }
> TAILQ_INSERT_TAIL(&devargs_list, devargs, next);
> return 0;
> diff --git a/lib/librte_eal/common/include/rte_bus.h
> b/lib/librte_eal/common/include/rte_bus.h
> index 6fb0834..331d954 100644
> --- a/lib/librte_eal/common/include/rte_bus.h
> +++ b/lib/librte_eal/common/include/rte_bus.h
> @@ -168,19 +168,19 @@ typedef int (*rte_bus_unplug_t)(struct rte_device *dev);
> typedef int (*rte_bus_parse_t)(const char *name, void *addr);
>
> /**
> - * Bus scan policies
> + * Bus probe policies
> */
> -enum rte_bus_scan_mode {
> - RTE_BUS_SCAN_UNDEFINED,
> - RTE_BUS_SCAN_WHITELIST,
> - RTE_BUS_SCAN_BLACKLIST,
> +enum rte_bus_probe_mode {
> + RTE_BUS_PROBE_UNDEFINED,
> + RTE_BUS_PROBE_WHITELIST,
> + RTE_BUS_PROBE_BLACKLIST,
> };
>
> /**
> * A structure used to configure bus operations.
> */
> struct rte_bus_conf {
> - enum rte_bus_scan_mode scan_mode; /**< Scan policy. */
> + enum rte_bus_probe_mode probe_mode; /**< Probe policy. */
> };
>
>
> --
> 2.1.4
>
--
Gaëtan Rivet
6WIND
Re: [dpdk-dev] [PATCH v6 1/2] doc: add generic compilation doc for all sample apps
> -Original Message- > From: Kovacevic, Marko > Sent: Wednesday, October 11, 2017 4:51 PM > To: Mcnamara, John > Cc: dev@dpdk.org; Kovacevic, Marko ; Herakliusz > Lipiec > Subject: [PATCH v6 1/2] doc: add generic compilation doc for all sample > apps > > From: Herakliusz Lipiec > > Moved duplicated, and occasionally outdated, doc sections from each > of the sample app guides chapters to a common chapter at the start. > > This reduces the duplication in the docs and provides a single > point of reference for compiling the sample apps. > > Signed-off-by: Herakliusz Lipiec > Signed-off-by: Marko Kovacevic > Reviewed-by: John McNamara Acked-by: John McNamara
Re: [dpdk-dev] [PATCH v6 2/2] doc: add new introduction to sample app guides
> -Original Message- > From: Kovacevic, Marko > Sent: Wednesday, October 11, 2017 4:51 PM > To: Mcnamara, John > Cc: dev@dpdk.org; Kovacevic, Marko > Subject: [PATCH v6 2/2] doc: add new introduction to sample app guides > > Add new Introduction Section into the sample app guides. > > Signed-off-by: Marko Kovacevic > Reviewed-by: John McNamara Acked-by: John McNamara
Re: [dpdk-dev] [PATCH V3 2/5] ethdev: add new eth_dev_ops function for mtr ops get
On 10/6/2017 8:15 PM, Cristian Dumitrescu wrote:
Following similar approach as rte_flow and rte_tm for modularity reasons,
the ops for the new rte_mtr API are retrieved through a new eth_dev_ops
function.
Signed-off-by: Cristian Dumitrescu
---
Changes in v3:
- None
Changes in v2:
-None
Changes in v1 (from RFC [1]):
- Removed ethdev API function to get the MTR ops, as it is not needed
(input from Thomas)
[1] RFC: http://www.dpdk.org/ml/archives/dev/2017-May/066888.html
lib/librte_ether/rte_ethdev.h | 6 ++
1 file changed, 6 insertions(+)
diff --git a/lib/librte_ether/rte_ethdev.h b/lib/librte_ether/rte_ethdev.h
index 99cdd54..a5802b0 100644
--- a/lib/librte_ether/rte_ethdev.h
+++ b/lib/librte_ether/rte_ethdev.h
@@ -1424,6 +1424,9 @@ typedef int (*eth_filter_ctrl_t)(struct rte_eth_dev *dev,
typedef int (*eth_tm_ops_get_t)(struct rte_eth_dev *dev, void *ops);
/**< @internal Get Traffic Management (TM) operations on an Ethernet device */
+typedef int (*eth_mtr_ops_get_t)(struct rte_eth_dev *dev, void *ops);
+/**< @internal Get Trafffic Metering and Policing (MTR) operations */
+
typedef int (*eth_get_dcb_info)(struct rte_eth_dev *dev,
struct rte_eth_dcb_info *dcb_info);
/**< @internal Get dcb information on an Ethernet device */
@@ -1548,6 +1551,9 @@ struct eth_dev_ops {
eth_tm_ops_get_t tm_ops_get;
/**< Get Traffic Management (TM) operations. */
+
+ eth_mtr_ops_get_t mtr_ops_get;
+ /**< Get Traffic Metering and Policing (MTR) operations. */
};
/**
Reviewed-by: Hemant Agrawal
Re: [dpdk-dev] [PATCH] examples/ip_reassembly: use pktmbuf to create pool
On 10/12/2017 4:08 PM, Ananyev, Konstantin wrote:
-Original Message-
From: Hemant Agrawal [mailto:hemant.agra...@nxp.com]
Sent: Wednesday, September 6, 2017 10:34 AM
To: dev@dpdk.org
Cc: Ananyev, Konstantin ; ashish.j...@nxp.com
Subject: [PATCH] examples/ip_reassembly: use pktmbuf to create pool
From: Ashish Jain
Use of rte_mempool_create() with the helper provided in
librte_mbuf: rte_pktmbuf_pool_create().
This is the preferred way to create a mbuf pool else
it may not work on implementation using the HW buffer pool
Signed-off-by: Ashish Jain
---
examples/ip_reassembly/main.c | 13 ++---
1 file changed, 6 insertions(+), 7 deletions(-)
diff --git a/examples/ip_reassembly/main.c b/examples/ip_reassembly/main.c
index e62636c..20caeb3 100644
--- a/examples/ip_reassembly/main.c
+++ b/examples/ip_reassembly/main.c
@@ -84,8 +84,7 @@
#define MAX_JUMBO_PKT_LEN 9600
#defineBUF_SIZERTE_MBUF_DEFAULT_DATAROOM
-#define MBUF_SIZE \
- (BUF_SIZE + sizeof(struct rte_mbuf) + RTE_PKTMBUF_HEADROOM)
+#defineMBUF_DATA_SIZE RTE_MBUF_DEFAULT_BUF_SIZE
#define NB_MBUF 8192
@@ -909,11 +908,11 @@ setup_queue_tbl(struct rx_queue *rxq, uint32_t lcore,
uint32_t queue)
snprintf(buf, sizeof(buf), "mbuf_pool_%u_%u", lcore, queue);
- if ((rxq->pool = rte_mempool_create(buf, nb_mbuf, MBUF_SIZE, 0,
- sizeof(struct rte_pktmbuf_pool_private),
- rte_pktmbuf_pool_init, NULL, rte_pktmbuf_init, NULL,
- socket, MEMPOOL_F_SP_PUT | MEMPOOL_F_SC_GET)) == NULL) {
- RTE_LOG(ERR, IP_RSMBL, "mempool_create(%s) failed", buf);
+ rxq->pool = rte_pktmbuf_pool_create(buf, nb_mbuf, 0, 0,
+ MBUF_DATA_SIZE, socket);
As we can't pass SC|SP anymore can we then setup mempool cache size to some
non-zero value?
Konstantin
#define MEMPOOL_CACHE_SIZE 256
do you think "256" will be ok?
+ if (rxq->pool == NULL) {
+ RTE_LOG(ERR, IP_RSMBL,
+ "rte_pktmbuf_pool_create(%s) failed", buf);
return -1;
}
--
2.7.4
[dpdk-dev] [PATCH v2 1/4] drivers: add queue parameter in event processing callback
For event dequeues for ethernet queues, a dpaa2_queue structure is
required to get the configuration for that ethernet queue. Based on
this configuration the rte_event is to be filled at the ingress side.
Signed-off-by: Nipun Gupta
---
drivers/bus/fslmc/portal/dpaa2_hw_pvt.h | 4
drivers/event/dpaa2/dpaa2_eventdev.c| 8 +++-
2 files changed, 11 insertions(+), 1 deletion(-)
diff --git a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
index 8f39cfb..e14a9a9 100644
--- a/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
+++ b/drivers/bus/fslmc/portal/dpaa2_hw_pvt.h
@@ -125,9 +125,12 @@ struct queue_storage_info_t {
int toggle;
};
+struct dpaa2_queue;
+
typedef void (dpaa2_queue_cb_dqrr_t)(struct qbman_swp *swp,
const struct qbman_fd *fd,
const struct qbman_result *dq,
+ struct dpaa2_queue *rxq,
struct rte_event *ev);
struct dpaa2_queue {
@@ -144,6 +147,7 @@ struct dpaa2_queue {
struct queue_storage_info_t *q_storage;
struct qbman_result *cscn;
};
+ struct rte_event ev;
dpaa2_queue_cb_dqrr_t *cb;
};
diff --git a/drivers/event/dpaa2/dpaa2_eventdev.c
b/drivers/event/dpaa2/dpaa2_eventdev.c
index 81286a8..8a6a3e3 100644
--- a/drivers/event/dpaa2/dpaa2_eventdev.c
+++ b/drivers/event/dpaa2/dpaa2_eventdev.c
@@ -189,10 +189,14 @@ static void dpaa2_eventdev_dequeue_wait(uint64_t
timeout_ticks)
static void dpaa2_eventdev_process_parallel(struct qbman_swp *swp,
const struct qbman_fd *fd,
const struct qbman_result *dq,
+ struct dpaa2_queue *rxq,
struct rte_event *ev)
{
struct rte_event *ev_temp =
(struct rte_event *)DPAA2_GET_FD_ADDR(fd);
+
+ RTE_SET_USED(rxq);
+
rte_memcpy(ev, ev_temp, sizeof(struct rte_event));
rte_free(ev_temp);
@@ -202,6 +206,7 @@ static void dpaa2_eventdev_process_parallel(struct
qbman_swp *swp,
static void dpaa2_eventdev_process_atomic(struct qbman_swp *swp,
const struct qbman_fd *fd,
const struct qbman_result *dq,
+ struct dpaa2_queue *rxq,
struct rte_event *ev)
{
struct rte_event *ev_temp =
@@ -209,6 +214,7 @@ static void dpaa2_eventdev_process_atomic(struct qbman_swp
*swp,
uint8_t dqrr_index = qbman_get_dqrr_idx(dq);
RTE_SET_USED(swp);
+ RTE_SET_USED(rxq);
rte_memcpy(ev, ev_temp, sizeof(struct rte_event));
rte_free(ev_temp);
@@ -265,7 +271,7 @@ static void dpaa2_eventdev_process_atomic(struct qbman_swp
*swp,
rxq = (struct dpaa2_queue *)qbman_result_DQ_fqd_ctx(dq);
if (rxq) {
- rxq->cb(swp, fd, dq, &ev[num_pkts]);
+ rxq->cb(swp, fd, dq, rxq, &ev[num_pkts]);
} else {
qbman_swp_dqrr_consume(swp, dq);
PMD_DRV_LOG(ERR, "Null Return VQ received\n");
--
1.9.1
