diff mbox series

[v3] selftests/mm: use asm volatile to not optimize mmap read variable

Message ID 20240606203619.677276-1-kernel@pankajraghav.com (mailing list archive)
State New
Headers show
Series [v3] selftests/mm: use asm volatile to not optimize mmap read variable | expand

Commit Message

Pankaj Raghav \(Samsung\) June 6, 2024, 8:36 p.m. UTC 
From: Pankaj Raghav <p.raghav@samsung.com>

create_pagecache_thp_and_fd() in split_huge_page_test.c used the
variable dummy to perform mmap read.

However, this test was skipped even on XFS which has large folio
support. The issue was compiler (gcc 13.2.0) was optimizing out the
dummy variable, therefore, not creating huge page in the page cache.

Use asm volatile() trick to force the compiler not to optimize out
the loop where we read from the mmaped addr. This is similar to what is
being done in other tests (cow.c, etc)

As the variable is now used in the asm statement, remove the unused
attribute.

Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
---
Changes since v2:
- Use the asm volatile trick to force the compiler to not optimize the
  read into dummy variable. (David)

 tools/testing/selftests/mm/split_huge_page_test.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)


base-commit: d97496ca23a2d4ee80b7302849404859d9058bcd

Comments

Zi Yan June 6, 2024, 8:52 p.m. UTC | #1 
On 6 Jun 2024, at 13:36, Pankaj Raghav (Samsung) wrote:

> From: Pankaj Raghav <p.raghav@samsung.com>
>
> create_pagecache_thp_and_fd() in split_huge_page_test.c used the
> variable dummy to perform mmap read.
>
> However, this test was skipped even on XFS which has large folio
> support. The issue was compiler (gcc 13.2.0) was optimizing out the
> dummy variable, therefore, not creating huge page in the page cache.
>
> Use asm volatile() trick to force the compiler not to optimize out
> the loop where we read from the mmaped addr. This is similar to what is
> being done in other tests (cow.c, etc)
>
> As the variable is now used in the asm statement, remove the unused
> attribute.
>
> Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
> ---
> Changes since v2:
> - Use the asm volatile trick to force the compiler to not optimize the
>   read into dummy variable. (David)
>
>  tools/testing/selftests/mm/split_huge_page_test.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
LGTM. Reviewed-by: Zi Yan <ziy@nvidia.com>

Best Regards,
Yan, Zi
David Hildenbrand June 6, 2024, 9:09 p.m. UTC | #2 
On 06.06.24 22:36, Pankaj Raghav (Samsung) wrote:
> From: Pankaj Raghav <p.raghav@samsung.com>
> 
> create_pagecache_thp_and_fd() in split_huge_page_test.c used the
> variable dummy to perform mmap read.
> 
> However, this test was skipped even on XFS which has large folio
> support. The issue was compiler (gcc 13.2.0) was optimizing out the
> dummy variable, therefore, not creating huge page in the page cache.
> 
> Use asm volatile() trick to force the compiler not to optimize out
> the loop where we read from the mmaped addr. This is similar to what is
> being done in other tests (cow.c, etc)
> 
> As the variable is now used in the asm statement, remove the unused
> attribute.
> 
> Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
> ---
> Changes since v2:
> - Use the asm volatile trick to force the compiler to not optimize the
>    read into dummy variable. (David)
> 
>   tools/testing/selftests/mm/split_huge_page_test.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/testing/selftests/mm/split_huge_page_test.c b/tools/testing/selftests/mm/split_huge_page_test.c
> index d3c7f5fb3e7b..e5e8dafc9d94 100644
> --- a/tools/testing/selftests/mm/split_huge_page_test.c
> +++ b/tools/testing/selftests/mm/split_huge_page_test.c
> @@ -300,7 +300,7 @@ int create_pagecache_thp_and_fd(const char *testfile, size_t fd_size, int *fd,
>   		char **addr)
>   {
>   	size_t i;
> -	int __attribute__((unused)) dummy = 0;
> +	int dummy = 0;
>   
>   	srand(time(NULL));
>   
> @@ -341,6 +341,7 @@ int create_pagecache_thp_and_fd(const char *testfile, size_t fd_size, int *fd,
>   
>   	for (size_t i = 0; i < fd_size; i++)
>   		dummy += *(*addr + i);
> +	asm volatile("" : "+r" (dummy));
>   
>   	if (!check_huge_file(*addr, fd_size / pmd_pagesize, pmd_pagesize)) {
>   		ksft_print_msg("No large pagecache folio generated, please provide a filesystem supporting large folio\n");
> 
> base-commit: d97496ca23a2d4ee80b7302849404859d9058bcd

Acked-by: David Hildenbrand <david@redhat.com>

I was wondering if we should factor that out into a mm-selftest helper 
function, but that can be done later.
Andrew Morton June 8, 2024, 4:39 a.m. UTC | #3 
On Thu,  6 Jun 2024 20:36:19 +0000 "Pankaj Raghav (Samsung)" <kernel@pankajraghav.com> wrote:

> From: Pankaj Raghav <p.raghav@samsung.com>
> 
> create_pagecache_thp_and_fd() in split_huge_page_test.c used the
> variable dummy to perform mmap read.
> 
> However, this test was skipped even on XFS which has large folio
> support. The issue was compiler (gcc 13.2.0) was optimizing out the
> dummy variable, therefore, not creating huge page in the page cache.
> 
> Use asm volatile() trick to force the compiler not to optimize out
> the loop where we read from the mmaped addr. This is similar to what is
> being done in other tests (cow.c, etc)
> 
> As the variable is now used in the asm statement, remove the unused
> attribute.
> 

What are the runtime effects of this change?  An inappropriate test
failure?  If so, shouldn't we fix 6.9.x kernels also?  And is
fc4d182316bd ("mm: huge_memory: enable debugfs to split huge pages to
any order") an appropriate Fixes: target?
Pankaj Raghav \(Samsung\) June 8, 2024, 10:22 a.m. UTC | #4 
On Fri, Jun 07, 2024 at 09:39:36PM -0700, Andrew Morton wrote:
> On Thu,  6 Jun 2024 20:36:19 +0000 "Pankaj Raghav (Samsung)" <kernel@pankajraghav.com> wrote:
> 
> > From: Pankaj Raghav <p.raghav@samsung.com>
> > 
> > create_pagecache_thp_and_fd() in split_huge_page_test.c used the
> > variable dummy to perform mmap read.
> > 
> > However, this test was skipped even on XFS which has large folio
> > support. The issue was compiler (gcc 13.2.0) was optimizing out the
> > dummy variable, therefore, not creating huge page in the page cache.
> > 
> > Use asm volatile() trick to force the compiler not to optimize out
> > the loop where we read from the mmaped addr. This is similar to what is
> > being done in other tests (cow.c, etc)
> > 
> > As the variable is now used in the asm statement, remove the unused
> > attribute.
> > 
> 
> What are the runtime effects of this change?  An inappropriate test
> failure?  If so, shouldn't we fix 6.9.x kernels also?  And is
> fc4d182316bd ("mm: huge_memory: enable debugfs to split huge pages to
> any order") an appropriate Fixes: target?

I mentioned it here in the commit message:

However, this test was skipped even on XFS which has large folio
support.

So the test just skip these cases and does not fail. That is why I
didn't use the Fixes tag. Let me know if it needs one.

--
Pankaj
diff mbox series

Patch

diff --git a/tools/testing/selftests/mm/split_huge_page_test.c b/tools/testing/selftests/mm/split_huge_page_test.c
index d3c7f5fb3e7b..e5e8dafc9d94 100644
--- a/tools/testing/selftests/mm/split_huge_page_test.c
+++ b/tools/testing/selftests/mm/split_huge_page_test.c
@@ -300,7 +300,7 @@  int create_pagecache_thp_and_fd(const char *testfile, size_t fd_size, int *fd,
 		char **addr)
 {
 	size_t i;
-	int __attribute__((unused)) dummy = 0;
+	int dummy = 0;
 
 	srand(time(NULL));
 
@@ -341,6 +341,7 @@  int create_pagecache_thp_and_fd(const char *testfile, size_t fd_size, int *fd,
 
 	for (size_t i = 0; i < fd_size; i++)
 		dummy += *(*addr + i);
+	asm volatile("" : "+r" (dummy));
 
 	if (!check_huge_file(*addr, fd_size / pmd_pagesize, pmd_pagesize)) {
 		ksft_print_msg("No large pagecache folio generated, please provide a filesystem supporting large folio\n");