[tip: x86/urgent] x86/cpu/amd: Fix workaround for erratum 1054

[tip: x86/urgent] x86/cpu/amd: Fix workaround for erratum 1054

[tip-bot2 for Sandipan Das]

The following commit has been merged into the x86/urgent branch of tip:

Commit-ID:     263e55949d8902a6a09bdb92a1ab6a3f67231abe
Gitweb:        https://git.kernel.org/tip/263e55949d8902a6a09bdb92a1ab6a3f67231abe
Author:        Sandipan Das <sandipan.das@amd.com>
AuthorDate:    Fri, 18 Apr 2025 11:49:40 +05:30
Committer:     Ingo Molnar <mingo@kernel.org>
CommitterDate: Fri, 18 Apr 2025 14:29:47 +02:00

x86/cpu/amd: Fix workaround for erratum 1054

Erratum 1054 affects AMD Zen processors that are a part of Family 17h
Models 00-2Fh and the workaround is to not set HWCR[IRPerfEn]. However,
when X86_FEATURE_ZEN1 was introduced, the condition to detect unaffected
processors was incorrectly changed in a way that the IRPerfEn bit gets
set only for unaffected Zen 1 processors.

Ensure that HWCR[IRPerfEn] is set for all unaffected processors. This
includes a subset of Zen 1 (Family 17h Models 30h and above) and all
later processors. Also clear X86_FEATURE_IRPERF on affected processors
so that the IRPerfCount register is not used by other entities like the
MSR PMU driver.

Fixes: 232afb557835 ("x86/CPU/AMD: Add X86_FEATURE_ZEN1")
Signed-off-by: Sandipan Das <sandipan.das@amd.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/caa057a9d6f8ad579e2f1abaa71efbd5bd4eaf6d.1744956467.git.sandipan.das@amd.com
---
 arch/x86/kernel/cpu/amd.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index a839ff5..2b36379 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -869,6 +869,16 @@ static void init_amd_zen1(struct cpuinfo_x86 *c)
 
 	pr_notice_once("AMD Zen1 DIV0 bug detected. Disable SMT for full protection.\n");
 	setup_force_cpu_bug(X86_BUG_DIV0);
+
+	/*
+	 * Turn off the Instructions Retired free counter on machines that are
+	 * susceptible to erratum #1054 "Instructions Retired Performance
+	 * Counter May Be Inaccurate".
+	 */
+	if (c->x86_model < 0x30) {
+		msr_clear_bit(MSR_K7_HWCR, MSR_K7_HWCR_IRPERF_EN_BIT);
+		clear_cpu_cap(c, X86_FEATURE_IRPERF);
+	}
 }
 
 static bool cpu_has_zenbleed_microcode(void)
@@ -1052,13 +1062,8 @@ static void init_amd(struct cpuinfo_x86 *c)
 	if (!cpu_feature_enabled(X86_FEATURE_XENPV))
 		set_cpu_bug(c, X86_BUG_SYSRET_SS_ATTRS);
 
-	/*
-	 * Turn on the Instructions Retired free counter on machines not
-	 * susceptible to erratum #1054 "Instructions Retired Performance
-	 * Counter May Be Inaccurate".
-	 */
-	if (cpu_has(c, X86_FEATURE_IRPERF) &&
-	    (boot_cpu_has(X86_FEATURE_ZEN1) && c->x86_model > 0x2f))
+	/* Enable the Instructions Retired free counter */
+	if (cpu_has(c, X86_FEATURE_IRPERF))
 		msr_set_bit(MSR_K7_HWCR, MSR_K7_HWCR_IRPERF_EN_BIT);
 
 	check_null_seg_clears_base(c);

Re: [tip: x86/urgent] x86/cpu/amd: Fix workaround for erratum 1054

[Borislav Petkov]

On Fri, Apr 18, 2025 at 02:11:44PM +0200, Ingo Molnar wrote:
> No, it doesn't really 'need' a stable tag, it has a Fixes tag already, 
> which gets processed by the -stable team.

Last time I asked Greg, he said they scan for those tags but it doesn't hurt
to Cc stable as it helps.

Greg?

> Also, the bug is old, 1.5 years old:
> 
>   Date: Sat, 2 Dec 2023 12:50:23 +0100
> 
> plus the erratum is a perf-counters information quality bug affecting 
> what appears to be a limited number of models, with the workaround 

No, the fix is needed because Zen2 and newer won't set
MSR_K7_HWCR_IRPERF_EN_BIT. It needs to go everywhere.

> likely incorporated in BIOS updates as well.

You can very much forget that argument. I have hard BIOS adoption data which
paints an abysmal picture. So NEVER EVER rely on BIOS to do anything.
Especially for Zen1 which is oooold in BIOS time.

> Leave it up to the -stable team whether they think it's severe enough to
> backport it?

No, they leave it to us section maintainers to decide AFAIK.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [tip: x86/urgent] x86/cpu/amd: Fix workaround for erratum 1054

[Ingo Molnar]

* Borislav Petkov <bp@alien8.de> wrote:

> > plus the erratum is a perf-counters information quality bug 
> > affecting what appears to be a limited number of models, with the 
> > workaround
> 
> No, the fix is needed because Zen2 and newer won't set 
> MSR_K7_HWCR_IRPERF_EN_BIT. It needs to go everywhere.

Fair enough - the current version already has Cc: stable:

  263e55949d89 x86/cpu/amd: Fix workaround for erratum 1054

  ...

  Fixes: 232afb557835 ("x86/CPU/AMD: Add X86_FEATURE_ZEN1")
  Signed-off-by: Sandipan Das <sandipan.das@amd.com>
  Signed-off-by: Ingo Molnar <mingo@kernel.org>
  Acked-by: Borislav Petkov <bp@alien8.de>
  Cc: stable@vger.kernel.org
  Link: https://lore.kernel.org/r/caa057a9d6f8ad579e2f1abaa71efbd5bd4eaf6d.1744956467.git.sandipan.das@amd.com

Thanks,

	Ingo

Re: [tip: x86/urgent] x86/cpu/amd: Fix workaround for erratum 1054

[Greg KH]

On Fri, Apr 18, 2025 at 02:37:13PM +0200, Borislav Petkov wrote:
> On Fri, Apr 18, 2025 at 02:11:44PM +0200, Ingo Molnar wrote:
> > No, it doesn't really 'need' a stable tag, it has a Fixes tag already, 
> > which gets processed by the -stable team.

NOOOOOO!!!!

> Last time I asked Greg, he said they scan for those tags but it doesn't hurt
> to Cc stable as it helps.
> 
> Greg?

Fixes: is a "best effort if we get around to it because a maintainer
forgot to put an actual cc: stable tag on it".

As the documentation has stated, since the start of the stable kernel
tree work, use a cc: stable tag if you want it to go to a stable tree.
Fixes came years later and we are forced to dig through them
occasionally because people forget.  But you do NOT get a FAILED email
if the commit does not apply to a stable tree, and sometimes we just
ignore them entirely if we are busy with other stuff.

So please ALWAYS use cc: stable@ on patches you know you want to be
applied to stable trees.  Use the Fixes: tag to tell us how far back to
backport them.  That's it.  Use both.

thanks,

greg k-h

Re: [tip: x86/urgent] x86/cpu/amd: Fix workaround for erratum 1054

[Ingo Molnar]

* Greg KH <gregkh@linuxfoundation.org> wrote:

> On Fri, Apr 18, 2025 at 02:37:13PM +0200, Borislav Petkov wrote:
> > On Fri, Apr 18, 2025 at 02:11:44PM +0200, Ingo Molnar wrote:
> > > No, it doesn't really 'need' a stable tag, it has a Fixes tag already, 
> > > which gets processed by the -stable team.
> 
> NOOOOOO!!!!

Noted!! :-)

> So please ALWAYS use cc: stable@ on patches you know you want to be 
> applied to stable trees.  Use the Fixes: tag to tell us how far back 
> to backport them.  That's it.  Use both.

Undertood!

Thanks,

	Ingo