x86/fpu: Delay instruction pointer fixup until after warning
authorDave Hansen <dave.hansen@linux.intel.com>
Tue, 24 Jun 2025 21:01:48 +0000 (14:01 -0700)
committerDave Hansen <dave.hansen@linux.intel.com>
Wed, 25 Jun 2025 23:28:06 +0000 (16:28 -0700)
commit1cec9ac2d071cfd2da562241aab0ef701355762a
treeef956d0af8d304f0dedf3a106152da08011752fb
parent8b05b3c988162ca117b3854ae7d497927b415299
x86/fpu: Delay instruction pointer fixup until after warning

Right now, if XRSTOR fails a console message like this is be printed:

Bad FPU state detected at restore_fpregs_from_fpstate+0x9a/0x170, reinitializing FPU registers.

However, the text location (...+0x9a in this case) is the instruction
*AFTER* the XRSTOR. The highlighted instruction in the "Code:" dump
also points one instruction late.

The reason is that the "fixup" moves RIP up to pass the bad XRSTOR and
keep on running after returning from the #GP handler. But it does this
fixup before warning.

The resulting warning output is nonsensical because it looks like the
non-FPU-related instruction is #GP'ing.

Do not fix up RIP until after printing the warning. Do this by using
the more generic and standard ex_handler_default().

Fixes: d5c8028b4788 ("x86/fpu: Reinitialize FPU registers if restoring FPU state fails")
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Chao Gao <chao.gao@intel.com>
Acked-by: Alison Schofield <alison.schofield@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc:stable@vger.kernel.org
Link: https://lore.kernel.org/all/20250624210148.97126F9E%40davehans-spike.ostc.intel.com
arch/x86/mm/extable.c