From: Leon Romanovsky
The kernel compiled with GCC 10.2.1 and KASAN together with GCOV enabled
produces the following splat while reloading modules. The very similar
trace was reported by Colin [1].
[ cut here ]
WARNING: CPU: 0 PID: 296 at mm/page_alloc.c:4859
From: Leon Romanovsky
Once GCOV fails to duplicate information, the following error is
printed:
gcov: could not save data for
'/home/leonro/src/kernel/drivers/infiniband/hw/mlx5/std_types.gcda' (out of
memory)
In the event of out-of-memory such prints are seen for almost every k
From: Leon Romanovsky
The filename is a const pointer, so use the proper string duplication
routine that takes into account const identifier.
Cc: Colin Ian King
Signed-off-by: Leon Romanovsky
---
kernel/gcov/gcc_4_7.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a
From: Leon Romanovsky
The kernel with KASAN and GCOV enabled generates the following splat
due to the situation that gcov_info can be both user and kernel pointer.
It is triggered by the memcpy() inside kmemdup(), so as a possible solution
let's copy fields man
From: Leon Romanovsky
Hi Linus,
Both Colin in Ubuntu [1] and I in FC 32 are having same kernel crashes
while GCOV is enabled. The reason to it that n_fuction variable that
should be provided by GCC is not initialized (or wrongly set).
This patch is based on the RFC [2] which I sent to gather
ile changed, 1 insertion(+), 1 deletion(-)
>
Fixes: 057729cb2347 ("RDMA/ocrdma: Remove driver QP state machine")
Thanks,
Reviewed-by: Leon Romanovsky
From: Leon Romanovsky
GCOV built with GCC 10 doesn't initialize n_function variable.
This produces different kernel panics as was seen by Colin in
Ubuntu [1] and me in FC 32 [2].
As a workaround, let's disable GCOV build for broken GCC 10 version.
[1] https://bugs.launchpad.net/ubun
t; 1 file changed, 2 deletions(-)
The patch is for rdma-rc and not for rdma-next.
Fixes: f86b31c6a28f ("RDMA/bnxt_re: Static NQ depth allocation")
Thanks,
Reviewed-by: Leon Romanovsky
From: Leon Romanovsky
Changelog:
v2:
* Added new patch to fix mlx4 failure on SR-IOV, it didn't have port set.
* Changed "RDMA/cma: Delete from restrack DB after successful destroy" patch.
v1:
* Fixed rebase error, deleted second assignment of qp_type.
* Rebased code on la
On Sat, Aug 29, 2020 at 5:12 PM Colin Ian King wrote:
>
> On 27/08/2020 14:39, Leon Romanovsky wrote:
> > From: Leon Romanovsky
> >
> > The kernel compiled with GCC 10.2.1 and KASAN together with GCOV enabled
> > produces the following splats while reloading modul
From: Leon Romanovsky
Changelog:
v1:
* Changed returned value in efa_destroy_ah() from EINVAL to EOPNOTSUPP
v0:
* https://lore.kernel.org/lkml/20200824103247.1088464-1-l...@kernel.org
-
Hi,
This series restores the
From: Leon Romanovsky
Changelog:
v1:
* Fixed rebase error, deleted second assignment of qp_type.
* Rebased code on latests rdma-next, the changes in cma.c caused to change
in patch "RDMA/cma: Delete from restrack DB after successful destroy".
* Dropped patch of port assignm
From: Leon Romanovsky
Changelog:
v2:
* Rebased on top of the 524d8ffd07f0
* Removed "udata" check in destroy flows
* Changed ib_free_cq to return early
* Used Jason's suggestion to implement "RDMA/mlx5: Issue FW command to destroy
SRQ on reentry" patch.
v1
*
("IB/mlx5: Enable subscription for device events over
> DEVX")
> Signed-off-by: Dinghao Liu
> ---
> drivers/infiniband/hw/mlx5/devx.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
Thanks,
Acked-by: Leon Romanovsky
On Thu, Oct 22, 2020 at 05:33:28PM -0700, Dave Ertman wrote:
<...>
> Dave Ertman (1):
> Add auxiliary bus support
We are in merge window now and both netdev and RDMA are closed for
submissions. So I'll send my mlx5 conversion patches once -rc1 will
be tagged.
However, It is important that thi
On Fri, Oct 23, 2020 at 08:56:10AM +0200, Greg KH wrote:
> On Fri, Oct 23, 2020 at 09:49:46AM +0300, Leon Romanovsky wrote:
> > On Thu, Oct 22, 2020 at 05:33:28PM -0700, Dave Ertman wrote:
> >
> > <...>
> >
> > > Dave Ertman (1):
> > > Add auxil
From: Leon Romanovsky
Hi,
This patch set converts mlx5 driver to use auxiliary bus [1].
In this series, we are connecting three subsystems (VDPA, netdev and
RDMA) through mlx5_core PCI driver. That driver is responsible to create
proper devices based on supported firmware.
First four patches
From: Leon Romanovsky
Change module registration logic to use auxiliary bus instead of custom
made mlx5 register interface.
Signed-off-by: Leon Romanovsky
---
drivers/net/ethernet/mellanox/mlx5/core/dev.c | 28 ++-
drivers/vdpa/mlx5/Makefile| 2 +-
drivers/vdpa/mlx5
From: Leon Romanovsky
The conversion to auxiliary bus solves long standing issue with
existing mlx5_ib<->mlx5_core coupling. It required to have both
modules in initramfs if one of them needed for the boot.
Signed-off-by: Leon Romanovsky
---
drivers/infiniband/hw/mlx5/ib_rep.c
From: Leon Romanovsky
Provide mlx5_core device instead of "priv" pointer while checking
eswith mode.
Signed-off-by: Leon Romanovsky
---
drivers/infiniband/hw/mlx5/counters.c | 7 ---
drivers/infiniband/hw/mlx5/ib_rep.c | 5 -
drivers/infiniba
From: Leon Romanovsky
Delete dead code.
Signed-off-by: Leon Romanovsky
---
drivers/infiniband/hw/mlx5/ib_rep.c | 31 +--
drivers/infiniband/hw/mlx5/ib_rep.h | 31 ---
drivers/net/ethernet/mellanox/mlx5/core/dev.c | 4 +--
3 files changed, 9
From: Leon Romanovsky
Remove exposed driver version as it was done in other drivers,
so module version will work correctly by displaying the kernel
version for which it is compiled.
And move mlx5_core module name to general include, so auxiliary drivers
will be able to use it as a basis for a
From: Leon Romanovsky
mlx5 firmware expects driver version in specific format X.X.X, so
make it always correct and based on real kernel version aligned with
the driver.
Fixes: 012e50e109fd ("net/mlx5: Set driver version into firmware")
Signed-off-by: Leon Romanovsky
---
drivers/ne
From: Leon Romanovsky
After conversion to use auxiliary bus, all custom device management is
not needed anymore, delete it.
Signed-off-by: Leon Romanovsky
---
drivers/net/ethernet/mellanox/mlx5/core/dev.c | 299 +-
drivers/net/ethernet/mellanox/mlx5/core/lag.c | 18
From: Leon Romanovsky
Users of mlx5_eswitch_get_vport() are required to check return value
prior to passing mlx5_vport further. Fix all the places to do not skip
that check.
Fixes: c7eddc6092b4 ("net/mlx5: E-switch, Move devlink eswitch ports closer to
eswitch")
Fixes: 5d9986a3947a
From: Leon Romanovsky
Create auxiliary devices under new virtual bus. This will replace
the custom-made mlx5 ->add()/->remove() interfaces and next patches
will fill the missing callback and remove the old interface logic.
The attachment of auxiliary drivers to the devices is possible in
From: Leon Romanovsky
Reuse auxiliary bus to perform device management of the
ethernet part of the mlx5 driver.
Signed-off-by: Leon Romanovsky
---
drivers/net/ethernet/mellanox/mlx5/core/dev.c | 74 ++
.../net/ethernet/mellanox/mlx5/core/en_main.c | 133 --
.../net
From: Leon Romanovsky
Move mlx5_vdpa IFC header file to the general include folder, so
mlx5_core will be able to reuse it to check if VDPA is supported
prior to creating an auxiliary device.
As part of this move, update the header file name to mlx5 general
naming scheme.
Reviewed-by: Parav
From: Leon Romanovsky
>From Jason:
The new common code does a better job finding large page sizes. Use it in
mlx5 for MRs.
This requires moving the MTT population for mailboxes and UMR over to
rdma_for_each_dma_block().
Thanks
Jason Gunthorpe (5):
RDMA/mlx5: Change mlx5_ib_populate_
From: Leon Romanovsky
>From Jason:
Move the remaining cases working with umems to use versions of
ib_umem_find_best_pgsz() tailored to the calculations the devices
requires.
Unlike a MR there is no IOVA, instead a page offset from the starting page
is possible, with various restricti
From: Leon Romanovsky
Addition to the IPoIB patches
https://lore.kernel.org/linux-rdma/20201026132904.1338526-1-l...@kernel.org/T/#u
Thanks
Meir Lichtinger (2):
IB/core: Add support for NDR link speed
IB/mlx5: Add support for NDR link speed
drivers/infiniband/core/sysfs.c | 4
From: Leon Romanovsky
Changelog:
v1:
* Renamed _mlx5_rescan_driver to be mlx5_rescan_driver_locked like in
other parts of the mlx5 driver.
* Renamed MLX5_INTERFACE_PROTOCOL_VDPA to tbe MLX5_INTERFACE_PROTOCOL_VNET as
a preparation to coming series from Eli C.
* Some small naming renames
From: Leon Romanovsky
Create auxiliary devices under new virtual bus. This will replace
the custom-made mlx5 ->add()/->remove() interfaces and next patches
will fill the missing callback and remove the old interface logic.
The attachment of auxiliary drivers to the devices is possible in
From: Leon Romanovsky
Remove exposed driver version as it was done in other drivers,
so module version will work correctly by displaying the kernel
version for which it is compiled.
And move mlx5_core module name to general include, so auxiliary drivers
will be able to use it as a basis for a
From: Leon Romanovsky
Users of mlx5_eswitch_get_vport() are required to check return value
prior to passing mlx5_vport further. Fix all the places to do not skip
that check.
Fixes: c7eddc6092b4 ("net/mlx5: E-switch, Move devlink eswitch ports closer to
eswitch")
Fixes: 5d9986a3947a
From: Leon Romanovsky
Move mlx5_vdpa IFC header file to the general include folder, so
mlx5_core will be able to reuse it to check if VDPA is supported
prior to creating an auxiliary device.
As part of this move, update the header file name to mlx5 general
naming scheme.
Reviewed-by: Parav
From: Leon Romanovsky
Delete dead code.
Signed-off-by: Leon Romanovsky
---
drivers/infiniband/hw/mlx5/ib_rep.c | 31 +++--
drivers/infiniband/hw/mlx5/ib_rep.h | 31 -
2 files changed, 7 insertions(+), 55 deletions(-)
diff --git a/drivers
From: Leon Romanovsky
After conversion to use auxiliary bus, all custom device management is
not needed anymore, delete it.
Signed-off-by: Leon Romanovsky
---
drivers/net/ethernet/mellanox/mlx5/core/dev.c | 288 ++
drivers/net/ethernet/mellanox/mlx5/core/lag.c | 18
From: Leon Romanovsky
mlx5 firmware expects driver version in specific format X.X.X, so
make it always correct and based on real kernel version aligned with
the driver.
Fixes: 012e50e109fd ("net/mlx5: Set driver version into firmware")
Signed-off-by: Leon Romanovsky
---
drivers/ne
From: Leon Romanovsky
Change module registration logic to use auxiliary bus instead of custom
made mlx5 register interface.
Signed-off-by: Leon Romanovsky
---
drivers/net/ethernet/mellanox/mlx5/core/dev.c | 28 ++-
drivers/vdpa/mlx5/Makefile| 2 +-
drivers/vdpa/mlx5
From: Leon Romanovsky
The conversion to auxiliary bus solves long standing issue with
existing mlx5_ib<->mlx5_core coupling. It required to have both
modules in initramfs if one of them needed for the boot.
Signed-off-by: Leon Romanovsky
---
drivers/infiniband/hw/mlx5/ib_rep.c
From: Leon Romanovsky
Provide mlx5_core device instead of "priv" pointer while checking
eswith mode.
Signed-off-by: Leon Romanovsky
---
drivers/infiniband/hw/mlx5/counters.c | 7 ---
drivers/infiniband/hw/mlx5/ib_rep.c | 5 -
drivers/infiniba
From: Leon Romanovsky
Reuse auxiliary bus to perform device management of the
ethernet part of the mlx5 driver.
Signed-off-by: Leon Romanovsky
---
drivers/net/ethernet/mellanox/mlx5/core/dev.c | 74 ++
.../net/ethernet/mellanox/mlx5/core/en_main.c | 133 --
.../net
On Mon, Nov 02, 2020 at 05:07:59AM +, Parav Pandit wrote:
>
>
> > From: Leon Romanovsky
> > Sent: Monday, November 2, 2020 1:46 AM
> >
> > From: Leon Romanovsky
> >
> > Remove exposed driver version as it was done in other drivers, so module
> &g
On Wed, Jul 08, 2020 at 01:28:35PM -0500, Gustavo A. R. Silva wrote:
> Hi Leon,
>
> On Wed, Jul 08, 2020 at 08:47:03AM +0300, Leon Romanovsky wrote:
> > On Tue, Jul 07, 2020 at 12:39:42PM -0500, Gustavo A. R. Silva wrote:
> > > Replace the existing /* fall through */ commen
On Thu, Nov 12, 2020 at 05:06:26PM +0800, Chen Zhou wrote:
> Fix to return a negative error code from the error handling case
> instead of 0 in function _ib_modify_qp(), as done elsewhere in this
> function.
>
> Fixes: 51aab12631dd ("RDMA/core: Get xmit slave for LAG")
> Reported-by: Hulk Robot
>
ridharan
> Co-developed-by: Fred Oh
> Signed-off-by: Fred Oh
> Co-developed-by: Leon Romanovsky
> Signed-off-by: Leon Romanovsky
> Reviewed-by: Pierre-Louis Bossart
> Reviewed-by: Shiraz Saleem
> Reviewed-by: Parav Pandit
> Reviewed-by: Dan Williams
> Signed-off-by: Da
From: Leon Romanovsky
Changelog:
v5:
* Reorder patches to postpone changes in rdma_restrack_add to be in next
series.
v4: https://lore.kernel.org/linux-rdma/20201104144008.3808124-1-l...@kernel.org/
* Rebased on latest for-upstream, all that time the patches were in
our regression and didn&#
On Tue, Nov 17, 2020 at 08:02:53AM +0100, Greg KH wrote:
> On Tue, Nov 17, 2020 at 07:30:00AM +0200, Leon Romanovsky wrote:
> > On Fri, Nov 13, 2020 at 08:18:50AM -0800, Dave Ertman wrote:
> > > Add support for the Auxiliary Bus, auxiliary_device and auxiliary_driver.
> >
On Tue, Nov 17, 2020 at 01:48:08PM +, Mark Brown wrote:
> On Tue, Nov 17, 2020 at 07:30:00AM +0200, Leon Romanovsky wrote:
> > On Fri, Nov 13, 2020 at 08:18:50AM -0800, Dave Ertman wrote:
>
> > > Add support for the Auxiliary Bus, auxiliary_device and auxiliary_driver.
>
From: Leon Romanovsky
v1: Added Jason's variant of first patch
v0: https://lore.kernel.org/lkml/20201012045600.418271-1-l...@kernel.org
Leon Romanovsky (2):
RDMA/core: Postpone uobject cleanup on failure till FD close
RDMA/core: Make FD destroy callback void
drivers/infiniband
.
> > > >
> > > > Co-developed-by: Kiran Patil
> > > > Signed-off-by: Kiran Patil
> > > > Co-developed-by: Ranjani Sridharan
> > > > Signed-off-by: Ranjani Sridharan
> > > > Co-developed-by: Fred Oh
> > > > Sign
On Sun, Nov 15, 2020 at 04:25:48PM +0800, Wang ShaoBo wrote:
> Fix to return the error code from hfi1_netdev_alloc() instaed of 0
> in hfi1_init_dd(), as done elsewhere in this function.
>
> Fixes: 4730f4a6c6b2 ("IB/hfi1: Activate the dummy netdev")
> Signed-off-by: Wang ShaoBo
> ---
> drivers/in
From: Leon Romanovsky
Changelog:
v1:
* Added patch for raw QP
* Fixed commit messages
v0: https://lore.kernel.org/lkml/20201026132635.1337663-1-l...@kernel.org
-
>From Jason:
Move the remaining cases working with umems to use versions of
ib_umem_find_best_p
From: Leon Romanovsky
Update mlx4 and mlx5 drivers to support querying AH for XRC QP types.
Thanks
Avihai Horon (2):
RDMA/mlx5: Enable querying AH for XRC QP types
RDMA/mlx4: Enable querying AH for XRC QP types
drivers/infiniband/hw/mlx4/qp.c | 4 +++-
drivers/infiniband/hw/mlx5/qp.c | 4
On Thu, Nov 05, 2020 at 01:19:09AM -0800, Dan Williams wrote:
> Some doc fixups, and minor code feedback. Otherwise looks good to me.
>
> On Thu, Oct 22, 2020 at 5:35 PM Dave Ertman wrote:
> >
<...>
> >
> > +config AUXILIARY_BUS
> > + bool
>
> tristate? Unless you need non-exported symbols
On Thu, Nov 05, 2020 at 09:12:51AM -0800, Dan Williams wrote:
> On Thu, Nov 5, 2020 at 1:47 AM Leon Romanovsky wrote:
> >
> > On Thu, Nov 05, 2020 at 01:19:09AM -0800, Dan Williams wrote:
> > > Some doc fixups, and minor code feedback. Otherwise looks good to me.
> >
ux-rdma ; Jason
> > Gunthorpe ; Doug Ledford ;
> > Netdev ; David Miller ;
> > Jakub Kicinski ; Greg KH ;
> > Ranjani Sridharan ; Pierre-Louis Bossart
> > ; Fred Oh ;
> > Parav Pandit ; Saleem, Shiraz
> > ; Patil, Kiran ; Linux
> > Kernel Mailing Lis
On Thu, Nov 05, 2020 at 01:32:40PM -0600, Pierre-Louis Bossart wrote:
>
> > > > +module_init(auxiliary_bus_init);
> > > > +module_exit(auxiliary_bus_exit);
> > > > +
> > > > +MODULE_LICENSE("GPL");
> > >
> > > Per above SPDX is v2 only, so...
> > >
> > > MODULE_LICENSE("GPL v2");
> > >
> >
> > adde
On Thu, Nov 05, 2020 at 05:09:48PM -0400, Jason Gunthorpe wrote:
> On Thu, Nov 05, 2020 at 12:59:20PM -0800, Saeed Mahameed wrote:
>
> > 2. you can always load a driver without its underlying device existed.
> > for example, you can load a pci device driver/module and it will load
> > and wait for
On Thu, Nov 05, 2020 at 04:36:57PM -0400, Jason Gunthorpe wrote:
> On Thu, Nov 05, 2020 at 12:31:52PM -0800, Saeed Mahameed wrote:
> > On Sun, 2020-11-01 at 22:15 +0200, Leon Romanovsky wrote:
> > > From: Leon Romanovsky
> > >
> > > Move mlx5_vdpa IFC header fi
On Mon, Jul 20, 2020 at 08:36:35AM -0600, Jonathan Corbet wrote:
> On Mon, 20 Jul 2020 17:07:16 +0300
> Leon Romanovsky wrote:
>
> > > Do *you* want to review that megapatch? The number of issues that have
> > > come up make it clear that these patches do, indeed, nee
From: Leon Romanovsky
Changelog
v1:
* Delete "pd" cleanup line
* Moved all cleaned feilds to be last in declaration list to improve
readability.
v0:
https://lore.kernel.org/lkml/20200719060319.77603-1-l...@kernel.org
---
On Mon, Jul 20, 2020 at 11:34:00PM +0200, Alexander A. Klimov wrote:
>
>
> Am 20.07.20 um 18:48 schrieb Leon Romanovsky:
> > On Mon, Jul 20, 2020 at 08:36:35AM -0600, Jonathan Corbet wrote:
> > > On Mon, 20 Jul 2020 17:07:16 +0300
> > > Leon Romanovsky wrote:
On Tue, Jul 21, 2020 at 08:43:00AM +0200, Alexander A. Klimov wrote:
>
>
> Am 21.07.20 um 08:05 schrieb Leon Romanovsky:
> > On Mon, Jul 20, 2020 at 11:34:00PM +0200, Alexander A. Klimov wrote:
> > >
> > >
> > > Am 20.07.20 um 18:48 schrieb Leon Romanov
From: Leon Romanovsky
Changelog:
v2:
* Fixed warning reported by kbuild where I wrote same assignment twice.
v1:
https://lore.kernel.org/lkml/20200716103956.1422139-1-l...@kernel.org
* v0 revealed old bug
* https://lore.kernel.org/linux-rdma/20200716102059.1420681-1-l...@kernel.org
that
From: Leon Romanovsky
Hi,
There are two change as were reported by kbuild. They are not important
enough to have Fixes line.
Thanks
Leon Romanovsky (2):
RDMA/uverbs: Remove redundant assignments
RDMA/uverbs: Silence shiftTooManyBitsSigned warning
drivers/infiniband/core/uverbs_cmd.c
On Sun, Jul 19, 2020 at 05:58:45PM +0200, Alexander A. Klimov wrote:
> Rationale:
> Reduces attack surface on kernel devs opening the links for MITM
> as HTTPS traffic is much harder to manipulate.
>
> Deterministic algorithm:
> For each file:
> If not .svg:
> For each line:
> If doesn'
On Mon, Jul 20, 2020 at 07:58:48AM -0600, Jonathan Corbet wrote:
> On Mon, 20 Jul 2020 07:56:26 +0300
> Leon Romanovsky wrote:
>
> > > Documentation/networking/rds.rst | 2 +-
> > > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > Why can't it be
On Tue, Jul 28, 2020 at 06:45:20PM +1000, Stephen Rothwell wrote:
> Hi all,
>
> Today's linux-next merge of the kspp tree got a conflict in:
>
> drivers/infiniband/core/uverbs_cmd.c
>
> between commit:
>
> 29f3fe1d6854 ("RDMA/uverbs: Remove redundant assignments")
>
> from the rdma tree and com
On Sun, Jul 05, 2020 at 11:45:28PM +0200, Alexander A. Klimov wrote:
> Rationale:
> Reduces attack surface on kernel devs opening the links for MITM
> as HTTPS traffic is much harder to manipulate.
>
> Deterministic algorithm:
> For each file:
> If not .svg:
> For each line:
> If doesn'
From: Leon Romanovsky
Changelog:
v2: Rewrote mlx5 part to avoid ib_xrcd allocation.
v1: https://lore.kernel.org/lkml/20200623111531.1227013-1-l...@kernel.org
Changed ib_dealloc_xrcd_user() do not iterate over tgt list, because
it is expected to be empty.
v0: https://lore.kernel.org/lkml
On Fri, Jul 24, 2020 at 10:41:12AM +0200, Pavel Machek wrote:
> This is user API, but likely noone uses it...? Fix it before it
> becomes problem.
>
> Signed-off-by: Pavel Machek (CIP)
>
>
> diff --git a/include/uapi/rdma/mlx5_user_ioctl_cmds.h
> b/include/uapi/rdma/mlx5_user_ioctl_cmds.h
> index
From: Leon Romanovsky
>From Jason:
This small series simplifies some of the RDMA CM state transitions
connected with DESTROYING states and in the process resolves a bug
discovered by syzkaller.
Thanks
Jason Gunthorpe (4):
RDMA/cma: Simplify DEVICE_REMOVAL for internal_id
RDMA/cma: Us
> ---
> Changes in v2:
> - Remove additional overlooked fall-through markings.
>
> drivers/infiniband/hw/hfi1/chip.c | 27 ++-
> 1 file changed, 14 insertions(+), 13 deletions(-)
>
Thanks,
Reviewed-by: Leon Romanovsky
From: Leon Romanovsky
Changelog:
v1:
* v0 revealed old bug
https://lore.kernel.org/linux-rdma/20200716102059.1420681-1-l...@kernel.org
that took a while to find.
* create_cq() was rewritten to make sure that uobj is properly initialized.
v0: https://lore.kernel.org/lkml
On Wed, Sep 09, 2020 at 03:06:07PM -0300, Jason Gunthorpe wrote:
> On Mon, Sep 07, 2020 at 03:09:12PM +0300, Leon Romanovsky wrote:
> > From: Leon Romanovsky
> >
> > Changelog:
> > v2:
> > * Rebased on top of the 524d8ffd07f0
> > * Removed "
with the Infiniband driver that allocates a single page for hold the
pages. For 1TB memory registration, the temporary buffer would consume only
4KB, instead of 2GB.
Signed-off-by: Maor Gottlieb
Signed-off-by: Leon Romanovsky
---
include/linux/scatterlist.h | 6 ++
lib/scatterlist.c
On Thu, Sep 10, 2020 at 02:52:01PM +0200, Peter Oberparleiter wrote:
> On 04.09.2020 17:58, Leon Romanovsky wrote:
> > GCOV built with GCC 10 doesn't initialize n_function variable.
> > This produces different kernel panics as was seen by Colin in
> > Ubuntu [1] and me
From: Leon Romanovsky
Hi,
This series mainly changes mlx4, mlx5, and mthca drivers to future
change of QP allocation scheme.
The rdmavt driver will be sent separately.
Thanks
Leon Romanovsky (10):
RDMA/mlx5: Embed GSI QP into general mlx5_ib QP
RDMA/mlx5: Reuse existing fields in parent
On Sun, Sep 20, 2020 at 10:05:39PM +0300, Oded Gabbay wrote:
> On Sun, Sep 20, 2020 at 11:47 AM Greg Kroah-Hartman
> wrote:
> >
> > On Sat, Sep 19, 2020 at 04:22:35PM -0300, Jason Gunthorpe wrote:
> > > On Sat, Sep 19, 2020 at 07:27:30PM +0200, Greg Kroah-Hartman wrote:
> > > > > It's probably her
On Mon, Sep 21, 2020 at 02:22:02PM +0300, Gal Pressman wrote:
> On 18/09/2020 18:28, Jason Gunthorpe wrote:
> > On Fri, Sep 18, 2020 at 06:15:52PM +0300, Oded Gabbay wrote:
> >
> >> I'm sorry, but you won't be able to convince me here that I need to
> >> "enslave" my entire code to RDMA, just becau
On Fri, Sep 18, 2020 at 11:23:18AM +0800, Liu Shixin wrote:
> On 2020/9/18 1:33, Leon Romanovsky wrote:
> > On Thu, Sep 17, 2020 at 02:24:51PM -0300, Jason Gunthorpe wrote:
> >> On Thu, Sep 17, 2020 at 08:05:11PM +0300, Leon Romanovsky wrote:
> >>> On Thu, Sep 17, 2
From: Leon Romanovsky
>From Avihai,
When an application is not using RDMA CM and if it is using multiple RDMA
devices with one or more RoCE ports, finding the right GID table entry is
a long process.
For example, with two RoCE dual-port devices in a system, when IP
failover is used between
.
* Add a function to chain SGE to the next page.
Signed-off-by: Maor Gottlieb
Signed-off-by: Leon Romanovsky
---
include/linux/scatterlist.h | 29 ++
lib/scatterlist.c | 110
2 files changed, 91 insertions(+), 48 deletions(-)
diff --git a
600.0MB
After512001.2MB
Signed-off-by: Maor Gottlieb
Signed-off-by: Leon Romanovsky
---
drivers/infiniband/Kconfig | 1 +
drivers/infiniband/core/umem.c | 90 +-
2 files changed, 12 insertions(+), 79 deletions(-)
diff --git a
more SGEs.
Since sg_alloc_table_from_pages allocate exactly the number of chunks,
therefore chunks are equal to the number of SG entries.
Signed-off-by: Maor Gottlieb
Signed-off-by: Leon Romanovsky
---
lib/scatterlist.c | 75 ---
1 file change
From: Leon Romanovsky
Changelog:
v1:
* Changed _sg_chain to be __sg_chain
* Added dependency on ARCH_NO_SG_CHAIN
* Removed struct sg_append
v0:
* https://lore.kernel.org/lkml/20200903121853.1145976-1-l...@kernel.org
On Sun, Sep 13, 2020 at 08:14:17AM +0200, Christophe JAILLET wrote:
> The wrappers in include/linux/pci-dma-compat.h should go away.
>
> The patch has been generated with the coccinelle script below and has been
> hand modified to replace GFP_ with a correct flag.
> It has been compile tested.
>
>
From: Leon Romanovsky
The various array_size functions use SIZE_MAX define, but missed limits.h
causes to failure to compile code that needs overflow.h.
In file included from drivers/infiniband/core/uverbs_std_types_device.c:6:
./include/linux/overflow.h: In function 'array_size':
From: Leon Romanovsky
Changelog:
v1:
* Moved git_type logic to cma_set_default_gid_type - Patch #2
* Changed signature of rdma_query_gid_table - Patch #3
* Changed i to be unsigned - Patch #3
* Fixed multiplication overflow - Patch #4
v0: https://lore.kernel.org/lkml/20200910142204.1309061-1
From: Leon Romanovsky
This flow has become crufty and confusing. Revise it so that the rules
on how UMR is used with MRs is much clearer and more correct.
Fixes a few minor bugs in ODP and rereg_mr where disallowed things were
not properly blocked.
Thanks
Jason Gunthorpe (5):
RDMA/mlx5
From: Leon Romanovsky
Based on:
https://lore.kernel.org/lkml/20200914112653.345244-1-l...@kernel.org/
---
>From Yishai:
This series improves ODP performance by moving to use the HMM API as of below.
From: Leon Romanovsky
Max's and Shravan's usernames were changed while @mellanox.com emails
were transferred to be @nvidia.com.
Fixes: f6da70d99c96 ("MAINTAINERS: Update Mellanox and Cumulus Network
addresses to new domain")
Signed-off-by: Leon Romanovsky
---
MAINTAIN
On Mon, Sep 14, 2020 at 05:15:15PM -0400, Peter Xu wrote:
> On Mon, Sep 14, 2020 at 02:34:36PM -0400, Peter Xu wrote:
> > On Mon, Sep 14, 2020 at 10:32:11AM -0700, Linus Torvalds wrote:
> > > On Mon, Sep 14, 2020 at 7:38 AM Jason Gunthorpe wrote:
> > > >
> > > > I don't have a detailed explanation
On Tue, Sep 15, 2020 at 01:16:00PM +0300, Necip Fazil Yildiran wrote:
> When RDMA_RXE is enabled and CRYPTO is disabled, it results in the
> following Kbuild warning:
>
> WARNING: unmet direct dependencies detected for CRYPTO_CRC32
> Depends on [n]: CRYPTO [=n]
> Selected by [y]:
> - RDMA_RXE
On Thu, Sep 24, 2020 at 12:02:43PM -0400, Michael S. Tsirkin wrote:
> On Thu, Sep 24, 2020 at 08:47:05AM -0700, Randy Dunlap wrote:
> > On 9/24/20 3:24 AM, Eli Cohen wrote:
> > > On Thu, Sep 24, 2020 at 05:30:55AM -0400, Michael S. Tsirkin wrote:
> > --- linux-next-20200917.orig/drivers/vdpa/K
function `mlx5_vdpa_handle_set_map':
> mr.c:(.text+0xf1d): undefined reference to `vhost_iotlb_itree_first'
>
> Signed-off-by: Randy Dunlap
> Cc: "Michael S. Tsirkin"
> Cc: Jason Wang
> Cc: virtualizat...@lists.linux-foundation.org
> Cc: Saeed Mahameed
>
On Fri, Sep 18, 2020 at 02:36:10PM +0300, Gal Pressman wrote:
> On 17/09/2020 20:18, Jason Gunthorpe wrote:
> > On Tue, Sep 15, 2020 at 11:46:58PM +0300, Oded Gabbay wrote:
> >> infrastructure for communication between multiple accelerators. Same
> >> as Nvidia uses NVlink, we use RDMA that we have
On Fri, Sep 18, 2020 at 02:56:09PM +0300, Oded Gabbay wrote:
> On Fri, Sep 18, 2020 at 2:52 PM Leon Romanovsky wrote:
> >
> > On Fri, Sep 18, 2020 at 02:36:10PM +0300, Gal Pressman wrote:
> > > On 17/09/2020 20:18, Jason Gunthorpe wrote:
> > > > On Tue, Se
801 - 900 of 1087 matches
Mail list logo
