On Sat, Apr 11, 2026 at 08:43:57PM +0200, Mark Kettenis wrote:
> > Date: Thu, 09 Apr 2026 13:55:43 +0200
> > From: Mark Kettenis <[email protected]>
> > 
> > > Date: Thu, 9 Apr 2026 10:30:17 +0200
> > > From: Claudio Jeker <[email protected]>
> > > 
> > > On Wed, Apr 08, 2026 at 11:45:23PM +0200, Mark Kettenis wrote:
> > > > > Date: Wed, 8 Apr 2026 19:24:56 +0200
> > > > > From: Jeremie Courreges-Anglas <[email protected]>
> > > > > 
> > > > > On Mon, Mar 16, 2026 at 01:19:36PM -0600, Theo de Raadt wrote:
> > > > > > Jeremie Courreges-Anglas <[email protected]> wrote:
> > > > > > 
> > > > > > > On Mon, Mar 16, 2026 at 12:18:05PM -0600, Theo de Raadt wrote:
> > > > > > > > I'm surprised at your proposal.
> > > > > > > > 
> > > > > > > > If this condition gets detected, why do you think it is fine to
> > > > > > > > continue?  A kernel data structure is seriously corrupted.
> > > > > > > 
> > > > > > > I'm not saying it's fine, sorry if my mail was too long to read. 
> > > > > > > ;)
> > > > > > > 
> > > > > > > 1. I'm not 100% sure the checks that trigger are correct, after 
> > > > > > > all
> > > > > > >   they're not using volatile reads.  Maaaaybe that's the bug but I
> > > > > > >   have no idea right now.
> > > > > > >   
> > > > > > > 2. Kurt had posted this on ports@ earlier, then on bugs@, so far 
> > > > > > > no
> > > > > > >   one has a fix and you recently tagged 7.9.  This diff is an 
> > > > > > > attempt
> > > > > > >   to make kmos' and users life easier before next release.  
> > > > > > > Obviously
> > > > > > >   everybody would be happier with a proper fix.  Maybe this 
> > > > > > > admittedly
> > > > > > >   incomplete fix will spark a discussion?
> > > > > > 
> > > > > > Maybe.
> > > > > > 
> > > > > > But you cannot delete that ddb enter.  You could replace it with a
> > > > > > panic.  If you continue to run after that printf, the system will 
> > > > > > just
> > > > > > crash in other unknown ways which are more difficult to debug.
> > > > > 
> > > > > We have proof that the system doesn't necessarily crash after that
> > > > > message is printed.  kmos tested the db_enter removal yesterday
> > > > > and confirmed that he got the message on the console without the
> > > > > system crashing.  Using the diff below, I got this today on my LDOM's
> > > > > console:
> > > > > 
> > > > > Apr  8 11:37:26 ports /bsd: ctx_free: context 1641 still active in 
> > > > > dmmu
> > > > > Apr  8 12:21:12 ports /bsd: ctx_free: context 7896 still active in 
> > > > > dmmu
> > > > > Apr  8 12:24:29 ports /bsd: ctx_free: context 3150 still active in 
> > > > > dmmu
> > > > > Apr  8 13:43:56 ports /bsd: ctx_free: context 4221 still active in 
> > > > > dmmu
> > > > > Apr  8 15:55:50 ports /bsd: ctx_free: context 1264 still active in 
> > > > > dmmu
> > > > > Apr  8 18:55:48 ports /bsd: ctx_free: context 5664 still active in 
> > > > > dmmu
> > > > > 
> > > > > The system is running many loops of perl subprocesses in an attempt to
> > > > > reproduce another bug:
> > > > > 
> > > > >   count=0; while perl t.pl; do count=$((count + 1)); done; echo $count
> > > > > 
> > > > > I have zero reason to believe that this is specific to perl.  eg it
> > > > > may happens when building rust which AFAIK doesn't use perl.
> > > > > 
> > > > > So I stand by my initial proposal (or the variant below).  I'm not
> > > > > happy either with our partial understanding of this issue, and if
> > > > > someone had a better fix, I'd be all for it.  BUT the db_enter() call
> > > > > in -current and next 7.9 has so far done more harm than good.
> > > > 
> > > > Sorry, but this is really bad.  It means stale TSB entries have been
> > > > left behind and may be re-used when the context is re-used.  And that
> > > > could lead to some serious memory corruption.
> > > 
> > > While this really indicates that we have yet another bug in the pmap code
> > > I think it is really hard to hit it. The TSB is horribly small and the
> > > system needs to cycle through 8k contexts before reuse. I think because of
> > > this the chance of this entry to remain in the TSB until reuse is very
> > > low.
> > 
> > This Murphy guy wants to have a word with you ;).
> > 
> > > That bug has probably been around for a very long time and was never
> > > noticed. Only because of this extra check busy systems hit this now.
> > > 
> > > > If we want to paper over this issue, we should at least invalidate the
> > > > stale TSB entry.  So something like:
> > > > 
> > > >         for (i = 0; i < TSBENTS; i++) {
> > > >                 tag = READ_ONCE(&tsb_dmmu[i].tag);
> > > >                 if (TSB_TAG_CTX(tag) == oldctx) {
> > > >                         atomic_cas_ulong(&tsb_dmmu[i].tag, tag, 
> > > > TSB_TAG_INVALID);
> > > >                         printf("ctx_free: context %d still active in 
> > > > dmmu\n", oldctx);
> > > >                 }
> > > >                 tag = READ_ONCE(&tsb_immu[i].tag);
> > > >                 if (TSB_TAG_CTX(tag) == oldctx) {
> > > >                         atomic_cas_ulong(&tsb_dmmu[i].tag, tag, 
> > > > TSB_TAG_INVALID);
> > > >                         printf("ctx_free: context %d still active in 
> > > > immu\n", oldctx);
> > > >                 }
> > > >         }
> > > 
> > > I agree that we should invalidate the entry. On top of this please extend
> > > the printf to show both the tag and data field of the entry.
> > 
> > Yeah, that wouldn't be a bad idea.  As long as the printing doesn't
> > turn this into a DoS.
> > 
> > I can probably come up with a proper (and tested) diff this weekend.
> > But I don't mind if somebody beats me to it.
> 
> Lightly tested by building some kernels and libc with make -j24.  I've
> never been able to trigger the db_enter() on my t5120, so I don't
> expect to ever hit this code myself.
> 
> ok?
> 
> 
> ? arch/sparc64/compile/GENERIC.MP/obj
> Index: arch/sparc64/sparc64/pmap.c
> ===================================================================
> RCS file: /cvs/src/sys/arch/sparc64/sparc64/pmap.c,v
> diff -u -p -r1.127 pmap.c
> --- arch/sparc64/sparc64/pmap.c       14 Dec 2025 12:37:22 -0000      1.127
> +++ arch/sparc64/sparc64/pmap.c       11 Apr 2026 18:43:27 -0000
> @@ -2579,6 +2579,7 @@ ctx_free(struct pmap *pm)
>  {
>       int oldctx;
>  #ifdef DIAGNOSTIC
> +     u_int64_t tag, data;
>       int i;
>  #endif
>       
> @@ -2597,10 +2598,25 @@ ctx_free(struct pmap *pm)
>               db_enter();
>       }
>       for (i = 0; i < TSBENTS; i++) {
> -             if (TSB_TAG_CTX(tsb_dmmu[i].tag) == oldctx ||
> -                 TSB_TAG_CTX(tsb_immu[i].tag) == oldctx) {
> -                     printf("ctx_free: context %d still active\n", oldctx);
> -                     db_enter();
> +             tag = READ_ONCE(tsb_dmmu[i].tag);
> +                if (TSB_TAG_CTX(tag) == oldctx) {
> +                     data = READ_ONCE(tsb_dmmu[i].data);
> +                     atomic_cas_ulong((unsigned long *)&tsb_dmmu[i].tag,
> +                         tag, TSB_TAG_INVALID);
> +                     printf("%s: context %d still active in DMMU TSB\n",
> +                         __func__, oldctx);
> +                     printf("%s: tag 0x%016llx data 0x%016llx\n",
> +                         __func__, tag, data);
> +             }
> +             tag = READ_ONCE(tsb_immu[i].tag);
> +             if (TSB_TAG_CTX(tag) == oldctx) {
> +                     data = READ_ONCE(tsb_immu[i].data);
> +                     atomic_cas_ulong((unsigned long *)&tsb_immu[i].tag,
> +                         tag, TSB_TAG_INVALID);
> +                     printf("%s: context %d still active in IMMU TSB\n",
> +                         __func__, oldctx);
> +                     printf("%s: tag 0x%016llx data 0x%016llx\n",
> +                         __func__, tag, data);
>               }
>       }
>  #endif

Running with this for some time now doing the perl FPU torture.

ctx_free: context 7952 still active in DMMU TSB
ctx_free: tag 0x1f10000000008845 data 0x8000000ed2b42490
ctx_free: context 994 still active in DMMU TSB
ctx_free: tag 0x03e200000001fd88 data 0x8000000ed2b42490

IMO this diff should be committed. It will help kmos@ to build ports.
OK claudio@
-- 
:wq Claudio

Reply via email to