linux-next: build failure after merge of the execve tree

linux-next: build failure after merge of the execve tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 909 bytes --]

Hi all,

After merging the execve tree, today's linux-next build (x86_64
allmodconfig) failed like this:

fs/binfmt_elf.c: In function 'load_elf_library':
fs/binfmt_elf.c:1323:37: error: unused variable 'len' [-Werror=unused-variable]
 1323 |         unsigned long elf_bss, bss, len;
      |                                     ^~~
fs/binfmt_elf.c:1323:32: error: unused variable 'bss' [-Werror=unused-variable]
 1323 |         unsigned long elf_bss, bss, len;
      |                                ^~~
fs/binfmt_elf.c:1323:23: error: unused variable 'elf_bss' [-Werror=unused-variable]
 1323 |         unsigned long elf_bss, bss, len;
      |                       ^~~~~~~
cc1: all warnings being treated as errors

Caused by commit

  f2a1e5fab9ae ("binfmt_elf: Use elf_load() for library")

I have used the execve tree from next-20230929 for today.

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: linux-next: build failure after merge of the execve tree

[Kees Cook]

On Tue, Oct 03, 2023 at 01:35:22PM +1100, Stephen Rothwell wrote:
> Hi all,
> 
> After merging the execve tree, today's linux-next build (x86_64
> allmodconfig) failed like this:
> 
> fs/binfmt_elf.c: In function 'load_elf_library':
> fs/binfmt_elf.c:1323:37: error: unused variable 'len' [-Werror=unused-variable]
>  1323 |         unsigned long elf_bss, bss, len;
>       |                                     ^~~
> fs/binfmt_elf.c:1323:32: error: unused variable 'bss' [-Werror=unused-variable]
>  1323 |         unsigned long elf_bss, bss, len;
>       |                                ^~~
> fs/binfmt_elf.c:1323:23: error: unused variable 'elf_bss' [-Werror=unused-variable]
>  1323 |         unsigned long elf_bss, bss, len;
>       |                       ^~~~~~~
> cc1: all warnings being treated as errors
> 
> Caused by commit
> 
>   f2a1e5fab9ae ("binfmt_elf: Use elf_load() for library")
> 
> I have used the execve tree from next-20230929 for today.

Whoops! Fixed now. Thanks. :)

-- 
Kees Cook

linux-next: build failure after merge of the execve tree

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 2193 bytes --]

Hi all,

After merging the execve tree, today's linux-next build (powerpc
ppc64_defconfig) failed like this:

fs/coredump.c: In function 'dump_vma_snapshot':
fs/coredump.c:1301:45: error: too few arguments to function 'get_dump_page'
 1301 |                         struct page *page = get_dump_page(addr);
      |                                             ^~~~~~~~~~~~~
In file included from fs/coredump.c:6:
include/linux/mm.h:2620:14: note: declared here
 2620 | struct page *get_dump_page(unsigned long addr, int *locked);
      |              ^~~~~~~~~~~~~

Caused by commit

  ff41385709f0 ("coredump: Only sort VMAs when truncating or core_sort_vma sysctl is set")

interacting with commit

  d6ff4c8f6522 ("fs: avoid mmap sem relocks when coredumping with many missing pages")

from the vfs-brauner tree.

I did the obvious (but probably wrong fix up below).

From: Stephen Rothwell <sfr@canb.auug.org.au>
Date: Mon, 24 Feb 2025 17:27:25 +1100
Subject: [PATCH] fix up for "coredump: Only sort VMAs when truncating or
 core_sort_vma sysctl is set"

interacting with commit

  d6ff4c8f6522 ("fs: avoid mmap sem relocks when coredumping with many missing pages")

from the vfs-brauner tree.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
---
 fs/coredump.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/fs/coredump.c b/fs/coredump.c
index 723c1ae9d2f4..eb75cd30862d 100644
--- a/fs/coredump.c
+++ b/fs/coredump.c
@@ -1243,6 +1243,7 @@ static bool dump_vma_snapshot(struct coredump_params *cprm)
 	VMA_ITERATOR(vmi, mm, 0);
 	int i = 0;
 	size_t sparse_vma_dump_size = 0;
+	int locked = 0;
 
 	/*
 	 * Once the stack expansion code is fixed to not change VMA bounds
@@ -1298,7 +1299,7 @@ static bool dump_vma_snapshot(struct coredump_params *cprm)
 
 		/* Subtract zero pages from the sparse_vma_dump_size. */
 		for (addr = m->start; addr < m->start + m->dump_size; addr += PAGE_SIZE) {
-			struct page *page = get_dump_page(addr);
+			struct page *page = get_dump_page(addr, &locked);
 
 			if (!page)
 				sparse_vma_dump_size -= PAGE_SIZE;
-- 
2.45.2

-- 
Cheers,
Stephen Rothwell

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]