Linux-mm Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Uladzislau Rezki <urezki@gmail.com>
To: Jill Ravaliya <jillravaliya@gmail.com>
Cc: akpm@linux-foundation.org, urezki@gmail.com, linux-mm@kvack.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH 2/2] selftests/mm: add test for vrealloc() shrink page freeing
Date: Sat, 23 May 2026 18:20:39 +0200	[thread overview]
Message-ID: <ahHT1wDkF4Dg3R7P@milan> (raw)
In-Reply-To: <20260507114854.41117-2-jillravaliya@gmail.com>

On Thu, May 07, 2026 at 05:18:54PM +0530, Jill Ravaliya wrote:
> Add a selftest that verifies vrealloc() frees physical pages
> when shrinking an allocation.
> 
> The test loads a kernel module that:
> 1. Allocates 10MB with vmalloc()
> 2. Touches all pages to force physical allocation
> 3. Shrinks to 2MB with vrealloc()
> 4. Verifies free page count increased after shrink
> 
> Without the fix, the test fails because no pages are freed.
> With the fix applied, the test passes confirming ~2048 pages
> are returned to the system after shrinking from 10MB to 2MB.
> 
> Tested on kernel 7.0.0 in QEMU.
> 
> Signed-off-by: Jill Ravaliya <jillravaliya@gmail.com>
> ---
>  tools/testing/selftests/mm/Makefile           |  5 ++
>  .../selftests/mm/vrealloc_shrink_test.c       | 65 +++++++++++++++++++
>  2 files changed, 70 insertions(+)
>  create mode 100644 tools/testing/selftests/mm/vrealloc_shrink_test.c
> 
> diff --git a/tools/testing/selftests/mm/Makefile b/tools/testing/selftests/mm/Makefile
> index cd24596cd..4eab7c76c 100644
> --- a/tools/testing/selftests/mm/Makefile
> +++ b/tools/testing/selftests/mm/Makefile
> @@ -104,6 +104,7 @@ TEST_GEN_FILES += hugetlb_dio
>  TEST_GEN_FILES += droppable
>  TEST_GEN_FILES += guard-regions
>  TEST_GEN_FILES += merge
> +TEST_GEN_FILES += vrealloc_shrink_test
>  TEST_GEN_FILES += rmap
>  TEST_GEN_FILES += folio_split_race_test
>  
> @@ -282,3 +283,7 @@ warn_missing_page_frag:
>  	echo "Warning: $(PAGE_FRAG_WARNING). page_frag test will be skipped." ; \
>  	echo
>  endif
> +
> +# vrealloc shrink test module
> +vrealloc_shrink_mod.ko: vrealloc_shrink_mod.c
> +	$(MAKE) -C $(KDIR) M=$(PWD) modules
> diff --git a/tools/testing/selftests/mm/vrealloc_shrink_test.c b/tools/testing/selftests/mm/vrealloc_shrink_test.c
> new file mode 100644
> index 000000000..cf4263074
> --- /dev/null
> +++ b/tools/testing/selftests/mm/vrealloc_shrink_test.c
> @@ -0,0 +1,65 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Test that vrealloc() frees physical pages when shrinking.
> + *
> + * vrealloc() shrink path previously zeroed unused memory and updated
> + * vm->requested_size, but never freed the physical pages backing the
> + * unused portion of the allocation. This test verifies the fix by
> + * loading a kernel module that directly measures nr_pages before and
> + * after vrealloc() shrink.
> + *
> + * Copyright (C) 2026 Jill Ravaliya
> + */
> +
> +#include <stdio.h>
> +#include <stdlib.h>
> +#include <string.h>
> +#include <unistd.h>
> +#include <fcntl.h>
> +#include "../kselftest.h"
> +
> +#define MODULE_NAME "vrealloc_shrink_mod"
> +#define DMESG_PASS  "vrealloc_shrink: PASS"
> +#define DMESG_FAIL  "vrealloc_shrink: FAIL"
> +
> +static int run_cmd(const char *cmd)
> +{
> +	return system(cmd);
> +}
> +
> +static int check_dmesg_for(const char *pattern)
> +{
> +	char cmd[256];
> +	snprintf(cmd, sizeof(cmd),
> +		 "dmesg | grep -q '%s'", pattern);
> +	return system(cmd) == 0;
> +}
> +
> +int main(void)
> +{
> +	ksft_print_header();
> +	ksft_set_plan(1);
> +
> +	/* Insert the test module */
> +	if (run_cmd("insmod " MODULE_NAME ".ko") != 0) {
> +		ksft_test_result_skip(
> +			"could not load %s.ko - is it built?\n",
> +			MODULE_NAME);
> +		ksft_finished();
> +	}
> +
> +	/* Check dmesg for pass/fail */
> +	if (check_dmesg_for(DMESG_PASS)) {
> +		ksft_test_result_pass(
> +			"vrealloc shrink frees physical pages\n");
> +	} else if (check_dmesg_for(DMESG_FAIL)) {
> +		ksft_test_result_fail(
> +			"vrealloc shrink did NOT free physical pages\n");
> +	} else {
> +		ksft_test_result_fail(
> +			"could not find test result in dmesg\n");
> +	}
> +
> +	run_cmd("rmmod " MODULE_NAME);
> +	ksft_finished();
> +}
> -- 
> 2.43.0
> 
We have a test case that covers vrealloc() functionality. See test_vmalloc.c
file.

--
Uladzislau Rezki


  parent reply	other threads:[~2026-05-23 16:20 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-07 11:48 [PATCH 1/2] mm/vmalloc: free unused pages when shrinking vrealloc() allocation Jill Ravaliya
2026-05-07 11:48 ` [PATCH 2/2] selftests/mm: add test for vrealloc() shrink page freeing Jill Ravaliya
2026-05-23 13:34   ` kernel test robot
2026-05-23 16:20   ` Uladzislau Rezki [this message]
2026-05-24  2:53     ` Jill Ravaliya
2026-05-07 17:17 ` [PATCH 1/2] mm/vmalloc: free unused pages when shrinking vrealloc() allocation Uladzislau Rezki
2026-05-07 20:26 ` [syzbot ci] " syzbot ci

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=ahHT1wDkF4Dg3R7P@milan \
    --to=urezki@gmail.com \
    --cc=akpm@linux-foundation.org \
    --cc=jillravaliya@gmail.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox