[next] x86: fs/hugetlbfs/inode.c:673:16: error: variable 'm_index' is uninitialized when used here [-Werror,-Wuninitialized]

[Naresh Kamboju]

The x86_64 and i386 clang builds failed due to these warnings / errors
on Linux next-20220825.

Reported-by: Linux Kernel Functional Testing <lk...@linaro.org>

fs/hugetlbfs/inode.c:673:16: error: variable 'm_index' is
uninitialized when used here [-Werror,-Wuninitialized]
m_start, m_index, truncate_op);
^~~~~~~
fs/hugetlbfs/inode.c:651:26: note: initialize the variable 'm_index'
to silence this warning
pgoff_t m_start, m_index;
^
= 0
1 error generated.
make[3]: *** [scripts/Makefile.build:250: fs/hugetlbfs/inode.o] Error 1
make[3]: Target '__build' not remade because of errors.
make[2]: *** [scripts/Makefile.build:525: fs/hugetlbfs] Error 2

Build link: https://builds.tuxbuild.com/2Dpvef6uVvzHMztbfQRmxJfPXZT/

Steps to reproduce:
-------------------
# To install tuxmake on your system globally:
# sudo pip3 install -U tuxmake

tuxmake --runtime podman --target-arch x86_64 --toolchain clang-14
--kconfig https://builds.tuxbuild.com/2Dpvef6uVvzHMztbfQRmxJfPXZT/config
LLVM=1 LLVM_IAS=1

--
Linaro LKFT
https://lkft.linaro.org

[Nathan Chancellor]

Hi Naresh,

On Thu, Aug 25, 2022 at 04:13:30PM +0530, Naresh Kamboju wrote:
> The x86_64 and i386 clang builds failed due to these warnings / errors
> on Linux next-20220825.
>
> Reported-by: Linux Kernel Functional Testing <lk...@linaro.org>
>
> fs/hugetlbfs/inode.c:673:16: error: variable 'm_index' is
> uninitialized when used here [-Werror,-Wuninitialized]
> m_start, m_index, truncate_op);
> ^~~~~~~
> fs/hugetlbfs/inode.c:651:26: note: initialize the variable 'm_index'
> to silence this warning
> pgoff_t m_start, m_index;
> ^
> = 0
> 1 error generated.

Thank you for the report! Commit fcc0d3d00d74 ("hugetlb: handle truncate
racing with page faults") introduced this warning; the Intel folks
caught this as well:

https://lore.kernel.org/202208251012...@intel.com/

As an aside, can you please update your contact for our mailing list to
ll...@lists.linux.dev? We do not use the Google Groups list anymore.

Cheers,
Nathan

[Mike Kravetz]

On 08/25/22 16:13, Naresh Kamboju wrote:
> The x86_64 and i386 clang builds failed due to these warnings / errors
> on Linux next-20220825.
>
> Reported-by: Linux Kernel Functional Testing <lk...@linaro.org>
>
> fs/hugetlbfs/inode.c:673:16: error: variable 'm_index' is
> uninitialized when used here [-Werror,-Wuninitialized]
> m_start, m_index, truncate_op);
> ^~~~~~~
> fs/hugetlbfs/inode.c:651:26: note: initialize the variable 'm_index'

Thanks! That really should be 'index' and not 'm_index'. The variable
m_index was used in a previous version of the code and should have been
deleted. Depending on the uninitialized value of m_index, this could
cause long delays as the called routine checks the range m_start to m_index.

Andrew, how would you like to address in your tree? The following patch
can be applied on top of the original (next commit fcc0d3d00d74 "hugetlb:
handle truncate racing with page faults"). Or, I could squash and
resend the original, or resend the entire series.


From 99aae359c3a744e6718045d2f24b55c84bed4d56 Mon Sep 17 00:00:00 2001
From: Mike Kravetz <mike.k...@oracle.com>
Date: Thu, 25 Aug 2022 09:49:08 -0700
Subject: [PATCH] hugetlb: fix/remove uninitialized variable in
remove_inode_hugepages

Code introduced for the routine remove_inode_hugepages by patch "hugetlb:
handle truncate racing with page faults", incorrectly uses a variable
m_index. This is a remnant from a previous version of the code when
under development. Use the correct variable 'index' and remove
'm_index' from the routine.

Signed-off-by: Mike Kravetz <mike.k...@oracle.com>
---
fs/hugetlbfs/inode.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c
index 52d9b390389b..2f1d6da1bafb 100644
--- a/fs/hugetlbfs/inode.c
+++ b/fs/hugetlbfs/inode.c
@@ -648,7 +648,7 @@ static void remove_inode_hugepages(struct inode *inode, loff_t lstart,
const pgoff_t start = lstart >> huge_page_shift(h);
const pgoff_t end = lend >> huge_page_shift(h);
pgoff_t m_end = lm_end >> huge_page_shift(h);
- pgoff_t m_start, m_index;
+ pgoff_t m_start;
struct folio_batch fbatch;
struct folio *folio;
pgoff_t next, index;
@@ -670,7 +670,7 @@ static void remove_inode_hugepages(struct inode *inode, loff_t lstart,
* due to a race with fault code.
*/
freed += fault_lock_inode_indicies(h, inode, mapping,
- m_start, m_index, truncate_op);
+ m_start, index, truncate_op);

/*
* Remove folio that was part of folio_batch.
--
2.37.1

[Andrew Morton]

On Thu, 25 Aug 2022 09:56:15 -0700 Mike Kravetz <mike.k...@oracle.com> wrote:

> Andrew, how would you like to address in your tree? The following patch
> can be applied on top of the original (next commit fcc0d3d00d74 "hugetlb:
> handle truncate racing with page faults"). Or, I could squash and
> resend the original, or resend the entire series.

This little fixlet is great, thanks.