On Monday 05 December 2016 03:54 PM, Jerin Jacob wrote:
On Mon, Dec 05, 2016 at 03:42:18PM +0530, Shreyansh Jain wrote:
Hello Jerin,
Hello Shreyansh,
On Sunday 04 December 2016 02:25 AM, Jerin Jacob wrote:
rte_eal_dev_init() is a misleading name.
It actually performs the driver->probe for vdev,
which is parallel to rte_eal_pci_probe.
Changed to rte_eal_vdev_probe for consistency and
moved the vdev specific probe to eal_common_vdev.c
Suggested-by: Shreyansh Jain <shreyansh.j...@nxp.com>
Signed-off-by: Jerin Jacob <jerin.ja...@caviumnetworks.com>
---
+int
+rte_eal_vdev_probe(void)
+{
+ struct rte_devargs *devargs;
+
+ /*
+ * Note that the dev_driver_list is populated here
+ * from calls made to rte_eal_driver_register from constructor functions
+ * embedded into PMD modules via the RTE_PMD_REGISTER_VDEV macro
+ */
+
+ /* call the init function for each virtual device */
+ TAILQ_FOREACH(devargs, &devargs_list, next) {
+
+ if (devargs->type != RTE_DEVTYPE_VIRTUAL)
+ continue;
+
+ if (rte_eal_vdev_init(devargs->virt.drv_name,
The situation now is:
rte_eal_init=>rte_eal_vdev_probe()=>rte_eal_vdev_init()=> driver->probe()
Even though I had suggested this, my intention was to completely do away
with rte_*_[v]dev_init as it is misleading.
rte_eal_init=>rte_eal_vdev_probe=>driver->probe()
IMO, We don't need to remove rte_eal_vdev_init() as it is an
application API that uses to create vdev driver instance.Moreover,
change and removing that name will result in ABI breakage.
grep -ri "rte_eal_vdev_init" app/
app/test/test_cryptodev.c: ret = rte_eal_vdev_init(
app/test/test_cryptodev.c: TEST_ASSERT_SUCCESS(rte_eal_vdev_init(
app/test/test_cryptodev.c: TEST_ASSERT_SUCCESS(rte_eal_vdev_init(
app/test/test_cryptodev.c: TEST_ASSERT_SUCCESS(rte_eal_vdev_init(
app/test/test_cryptodev.c: TEST_ASSERT_SUCCESS(rte_eal_vdev_init(
app/test/test_cryptodev.c: int dev_id = rte_eal_vdev_init(
app/test/test_cryptodev.c: ret = rte_eal_vdev_init(
app/test/test_cryptodev_perf.c: ret = rte_eal_vdev_init(
app/test/test_cryptodev_perf.c: ret = rte_eal_vdev_init(
app/test/test_cryptodev_perf.c: ret = rte_eal_vdev_init(
app/test/test_cryptodev_perf.c: ret = rte_eal_vdev_init(
Got it.
Have you noticed patches from Ben which actually merges init and probe
all together? [1]. It is for PCI right now (and that too would break
ABIs, I am assuming).
[1] http://dpdk.org/dev/patchwork/patch/17206/
should be the ideal order, IMO.
Apologies, I was not completely clear then.
+ devargs->args)) {
+ RTE_LOG(ERR, EAL, "failed to initialize %s device\n",
+ devargs->virt.drv_name);
+ return -1;
+ }
+ }
+
+ return 0;
+}
diff --git a/lib/librte_eal/common/include/rte_dev.h
b/lib/librte_eal/common/include/rte_dev.h
index 8840380..146f505 100644
--- a/lib/librte_eal/common/include/rte_dev.h
+++ b/lib/librte_eal/common/include/rte_dev.h
@@ -171,9 +171,9 @@ void rte_eal_driver_register(struct rte_driver *driver);
void rte_eal_driver_unregister(struct rte_driver *driver);
/**
- * Initalize all the registered drivers in this process
+ * Probe all the registered vdev drivers in this process
*/
-int rte_eal_dev_init(void);
+int rte_eal_vdev_probe(void);
/**
* Initialize a driver specified by name.
diff --git a/lib/librte_eal/linuxapp/eal/eal.c
b/lib/librte_eal/linuxapp/eal/eal.c
index 16dd5b9..faf75cf 100644
--- a/lib/librte_eal/linuxapp/eal/eal.c
+++ b/lib/librte_eal/linuxapp/eal/eal.c
@@ -884,8 +884,8 @@ rte_eal_init(int argc, char **argv)
if (rte_eal_pci_probe())
rte_panic("Cannot probe PCI\n");
- if (rte_eal_dev_init() < 0)
- rte_panic("Cannot init pmd devices\n");
+ if (rte_eal_vdev_probe() < 0)
+ rte_panic("Cannot probe vdev drivers\n");
rte_eal_mcfg_complete();
diff --git a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
index 83721ba..67fc95b 100644
--- a/lib/librte_eal/linuxapp/eal/rte_eal_version.map
+++ b/lib/librte_eal/linuxapp/eal/rte_eal_version.map
@@ -22,7 +22,7 @@ DPDK_2.0 {
rte_dump_tailq;
rte_eal_alarm_cancel;
rte_eal_alarm_set;
- rte_eal_dev_init;
+ rte_eal_vdev_probe;
rte_eal_devargs_add;
rte_eal_devargs_dump;
rte_eal_devargs_type_count;
