Re: [PATCH 1/7] x86/vdso: Respect COMPAT_32BIT_TIME

["Arnd Bergmann" <arnd@arndb.de>]

On Tue, Mar 3, 2026, at 22:00, H. Peter Anvin wrote:
> On March 3, 2026 12:50:33 PM PST, Arnd Bergmann <arnd@arndb.de> wrote:
>>On Tue, Mar 3, 2026, at 19:11, H. Peter Anvin wrote:
>>> A bigger question is whether or not we should omit these from the vDSO
>>> completely (potentially causing link failures) or replace them with stubs
>>> returning -ENOSYS.
>>
>>I see no harm in keeping gettimeofday() in the vdso when
>>COMPAT_32BIT_TIME is turned on, as existing code will call it
>>no matter whether it's in the vdso or the syscall.
>
> That wasn't the point. The point was what kind of error behavior we want.

I see. I would argue here that a link failure is actually the best
case then: the entire point of CONFIG_COMPAT_32BIT_TIME=n is to
enforce the fail-early case, as opposed to silent data corruption
after (now) 12 years of operation.

I don't think there is an actual link failure here, as dl_vdso_vsym()
resolves it at runtime by glibc (when using time32), while musl
doesn't use __vdso_gettimeofday at all.

      Arnd