Re: [regression] Re: [PATCH 2/3] futex: Sanitize cmpxchg_futex_value_locked API

[Jonathan Nieder <jrnieder@gmail.com>]

Luck, Tony wrote:

>> It doesn't look like the return value (r8) is actually being set beyond
>> initialized to 0. If there is some ia64 instruction that modifies it, GCC
>> doesn't know about it from the inline assembly (r8 doesn't appear in the
>> inputs/outputs list). From looking at the x86 version (agh, inline asm is
>> hard to parse), it does modify the return value based on whether the
>> comparison was a success or not, and the return value is certainly used by
>> the callers.
>
> The commit comment for the change makes it sound like the return value
> is an error code (-ENOSYS if the function isn't implemented/configured;
> -EFAULT if the user address is bogus) - or zero if nothing bad happened.

Yes, that's right.

> Not "the comparison was a success or not".
>
> What's the real answer? The ia64 code is returning 0 regardless of whether the
> compare/exchange found the old value or not.  Is this a bad assumption?

No, I think something else is wrong, though I don't know what it would
be.

Émeric, was the bisection result reproducible?  E.g., if you try
building 37a9d912b24f and 37a9d912b24f^ again, does the former
consistently produce and the latter consistently not produce a crashy
system?