Re: git: 753a166bdeb3 - main - imgact_elf: Fix uninitialized variable use in note_procstat_auxv

On 25 Apr 2026, at 15:46, Jessica Clarke <jrtc27@freebsd.org> wrote:
> 
> On 25 Apr 2026, at 14:42, Dimitry Andric <dim@FreeBSD.org> wrote:
> 
>> The branch main has been updated by dim:
>> 
>> URL: https://cgit.FreeBSD.org/src/commit/?id=753a166bdeb3aeba02fd9678e7360f0929007368
>> 
>> commit 753a166bdeb3aeba02fd9678e7360f0929007368
>> Author:     Alex Richardson <arichardson@FreeBSD.org>
>> AuthorDate: 2025-09-15 06:27:12 +0000
>> Commit:     Dimitry Andric <dim@FreeBSD.org>
>> CommitDate: 2026-04-25 13:42:16 +0000
>> 
>>   imgact_elf: Fix uninitialized variable use in note_procstat_auxv
>> 
>>   Found building with latest clang
>> 
>>   MFC after:      3 days
>> ---
>> sys/kern/imgact_elf.c | 9 ++++++---
>> 1 file changed, 6 insertions(+), 3 deletions(-)
>> 
>> diff --git a/sys/kern/imgact_elf.c b/sys/kern/imgact_elf.c
>> index af0841c75549..3c5fbe4df342 100644
>> --- a/sys/kern/imgact_elf.c
>> +++ b/sys/kern/imgact_elf.c
>> @@ -2712,13 +2712,16 @@ __elfN(note_procstat_auxv)(void *arg, struct sbuf *sb, size_t *sizep)
>> struct proc *p;
>> size_t size;
>> int structsize;
>> -
>> +#if defined(COMPAT_FREEBSD32) && __ELF_WORD_SIZE == 32
>> + structsize = sizeof(Elf32_Auxinfo);
> 
> Elf_Auxinfo *is* Elf32_Auxinfo for __ELF_WORD_SIZE == 32?
> 
>> +#else
>> + structsize = sizeof(Elf_Auxinfo);
>> +#endif
>> p = arg;
>> structsize = sizeof(Elf_Auxinfo);
> 
> But it’s initialised right here? So that assignment is dead?
> 
>> if (sb == NULL) {
>> size = 0;
>> - sb = sbuf_new(NULL, NULL, AT_COUNT * sizeof(Elf_Auxinfo),
>> -    SBUF_FIXEDLEN);
>> + sb = sbuf_new(NULL, NULL, AT_COUNT * structsize, SBUF_FIXEDLEN);
> 
> These are equivalent? Though this one at least I see an argument for.
> 
>> sbuf_set_drain(sb, sbuf_count_drain, &size);
>> sbuf_bcat(sb, &structsize, sizeof(structsize));
>> PHOLD(p);
>> 
> 
> I really don’t understand this change at all. I think all but the final
> part should be reverted

I've only committed this because it's part of the llvm-21-update tree, and on behalf of Alex. I think this was to suppress a bunch of warnings, and I'm happing to fix it post-import, if there are no further errors.

-Dimitry