[PATCH net] selftests: drv-net: devmem: Skip when hugetlbfs isn't supported

[PATCH net] selftests: drv-net: devmem: Skip when hugetlbfs isn't supported

[Maxime Chevallier]

The devmem selftest requires accessing /proc/sys/vm/nr_hugepages,
which is only available when hugetlbfs is supported. This isn't the
case on all CPU architectures, for example it isn't supported on ARM
32-bits on cores that don't support LPAE.

On these devices, the devmem kselftest doesn't skip gracefully, it hits
an exception while trying to open the nr_hugepages file :

  # Exception|     _reserve_hugepages()
  # Exception|     ~~~~~~~~~~~~~~~~~~^^
  # Exception|   File "/usr/lib/kselftests/drivers/net/hw/devmem_lib.py", line 36, in _reserve_hugepages
  # Exception|     with open(NR_HUGEPAGES_FILE, 'r+', encoding='utf-8') as f:
  # Exception|          ~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  # Exception| FileNotFoundError: [Errno 2] No such file or directory: '/proc/sys/vm/nr_hugepages'

Let's check if the file exists before trying to open it, and return
immediately if we can't.

The _reserve_hugepages() runs as part of the "require_devmem()", that
skips if devmem itself isn't supported. Before the blamed commit, this
check was done prior to accessing nr_hugepages, so the problem wasn't
showing.

This has been found by trying to run the selftest on stmmac, on an
Altera CycloneV SoC (Cortex A9, doesn't support hugepages).

Fixes: 8ac4255c1e0c ("selftests/net: devmem.py: add check_rx_large_niov")
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
---
 tools/testing/selftests/drivers/net/hw/devmem_lib.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/tools/testing/selftests/drivers/net/hw/devmem_lib.py b/tools/testing/selftests/drivers/net/hw/devmem_lib.py
index 3554954a6691..24637fb180ca 100644
--- a/tools/testing/selftests/drivers/net/hw/devmem_lib.py
+++ b/tools/testing/selftests/drivers/net/hw/devmem_lib.py
@@ -33,6 +33,8 @@ def _restore_nr_hugepages(nr_hugepages):
 
 def _reserve_hugepages(want=64):
     """Raise nr_hugepages to @want and arrange for it to be restored."""
+    if not os.path.exists(NR_HUGEPAGES_FILE):
+        return
     with open(NR_HUGEPAGES_FILE, 'r+', encoding='utf-8') as f:
         nr_hugepages = int(f.read().strip())
         if nr_hugepages >= want:
-- 
2.56.0

Re: [PATCH net] selftests: drv-net: devmem: Skip when hugetlbfs isn't supported

[Björn Töpel]

Maxime Chevallier <maxime.chevallier@bootlin.com> writes:

> The devmem selftest requires accessing /proc/sys/vm/nr_hugepages,
> which is only available when hugetlbfs is supported. This isn't the
> case on all CPU architectures, for example it isn't supported on ARM
> 32-bits on cores that don't support LPAE.
>
> On these devices, the devmem kselftest doesn't skip gracefully, it hits
> an exception while trying to open the nr_hugepages file :
>
>   # Exception|     _reserve_hugepages()
>   # Exception|     ~~~~~~~~~~~~~~~~~~^^
>   # Exception|   File "/usr/lib/kselftests/drivers/net/hw/devmem_lib.py", line 36, in _reserve_hugepages
>   # Exception|     with open(NR_HUGEPAGES_FILE, 'r+', encoding='utf-8') as f:
>   # Exception|          ~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>   # Exception| FileNotFoundError: [Errno 2] No such file or directory: '/proc/sys/vm/nr_hugepages'
>
> Let's check if the file exists before trying to open it, and return
> immediately if we can't.
>
> The _reserve_hugepages() runs as part of the "require_devmem()", that
> skips if devmem itself isn't supported. Before the blamed commit, this
> check was done prior to accessing nr_hugepages, so the problem wasn't
> showing.
>
> This has been found by trying to run the selftest on stmmac, on an
> Altera CycloneV SoC (Cortex A9, doesn't support hugepages).
>
> Fixes: 8ac4255c1e0c ("selftests/net: devmem.py: add check_rx_large_niov")
> Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
> ---
>  tools/testing/selftests/drivers/net/hw/devmem_lib.py | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/tools/testing/selftests/drivers/net/hw/devmem_lib.py b/tools/testing/selftests/drivers/net/hw/devmem_lib.py
> index 3554954a6691..24637fb180ca 100644
> --- a/tools/testing/selftests/drivers/net/hw/devmem_lib.py
> +++ b/tools/testing/selftests/drivers/net/hw/devmem_lib.py
> @@ -33,6 +33,8 @@ def _restore_nr_hugepages(nr_hugepages):
>  
>  def _reserve_hugepages(want=64):
>      """Raise nr_hugepages to @want and arrange for it to be restored."""
> +    if not os.path.exists(NR_HUGEPAGES_FILE):
> +        return

Maybe raise a KsftSkipEx() would be more robust? No biggie...

Reviewed-by: Björn Töpel <bjorn@kernel.org>

Re: [PATCH net] selftests: drv-net: devmem: Skip when hugetlbfs isn't supported

[Stanislav Fomichev]

On 10/21, Maxime Chevallier wrote:
> The devmem selftest requires accessing /proc/sys/vm/nr_hugepages,
> which is only available when hugetlbfs is supported. This isn't the
> case on all CPU architectures, for example it isn't supported on ARM
> 32-bits on cores that don't support LPAE.
> 
> On these devices, the devmem kselftest doesn't skip gracefully, it hits
> an exception while trying to open the nr_hugepages file :
> 
>   # Exception|     _reserve_hugepages()
>   # Exception|     ~~~~~~~~~~~~~~~~~~^^
>   # Exception|   File "/usr/lib/kselftests/drivers/net/hw/devmem_lib.py", line 36, in _reserve_hugepages
>   # Exception|     with open(NR_HUGEPAGES_FILE, 'r+', encoding='utf-8') as f:
>   # Exception|          ~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>   # Exception| FileNotFoundError: [Errno 2] No such file or directory: '/proc/sys/vm/nr_hugepages'
> 
> Let's check if the file exists before trying to open it, and return
> immediately if we can't.
> 
> The _reserve_hugepages() runs as part of the "require_devmem()", that
> skips if devmem itself isn't supported. Before the blamed commit, this
> check was done prior to accessing nr_hugepages, so the problem wasn't
> showing.
> 
> This has been found by trying to run the selftest on stmmac, on an
> Altera CycloneV SoC (Cortex A9, doesn't support hugepages).
> 
> Fixes: 8ac4255c1e0c ("selftests/net: devmem.py: add check_rx_large_niov")
> Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>

Acked-by: Stanislav Fomichev <sdf@fomichev.me>

Re: [PATCH net] selftests: drv-net: devmem: Skip when hugetlbfs isn't supported

[Jakub Kicinski]

On Wed, 21 Oct 2026 11:08:51 +0200 Maxime Chevallier wrote:
> The devmem selftest requires accessing /proc/sys/vm/nr_hugepages,
> which is only available when hugetlbfs is supported. This isn't the
> case on all CPU architectures, for example it isn't supported on ARM
> 32-bits on cores that don't support LPAE.
> 
> On these devices, the devmem kselftest doesn't skip gracefully

We don't require graceful skipping. If anything IMO hiding the stack
traces isn't great.

If the check is extraneous and makes us fail some tests which don't
actually require huge pages please make that clear in the commit
message. Otherwise the net benefit of the changes is unclear.
-- 
pw-bot: cr