NetBSD Problem Report #60479

From www@netbsd.org  Wed Jul 22 20:50:43 2026
Return-Path: <www@netbsd.org>
Received: from mail.netbsd.org (mail.netbsd.org [199.233.217.200])
	(using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits)
	 key-exchange X25519 server-signature RSA-PSS (2048 bits)
	 client-signature RSA-PSS (2048 bits))
	(Client CN "mail.netbsd.org", Issuer "YR2" (not verified))
	by mollari.NetBSD.org (Postfix) with ESMTPS id B00361A923B
	for <gnats-bugs@gnats.NetBSD.org>; Wed, 22 Jul 2026 20:50:43 +0000 (UTC)
Message-Id: <20260722205042.DAE9C1A923C@mollari.NetBSD.org>
Date: Wed, 22 Jul 2026 20:50:42 +0000 (UTC)
From: campbell+netbsd@mumble.net
Reply-To: campbell+netbsd@mumble.net
To: gnats-bugs@NetBSD.org
Subject: alpha fails to save/restore fpu registers on signal delivery
X-Send-Pr-Version: www-1.0
X-From4GNATS: "campbell+netbsd@mumble.net via gnats" <gnats-admin@NetBSD.org>

>Number:         60479
>Category:       port-alpha
>Synopsis:       alpha fails to save/restore fpu registers on signal delivery
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    thorpej
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Jul 22 20:55:00 +0000 2026
>Last-Modified:  Sat Jul 25 04:40:01 +0000 2026
>Originator:     Taylor R Campbell
>Release:        
>Organization:
The AlphaBSD Fpundation, Inc.
>Environment:
>Description:

	FPU registers are not saved and restored on signal delivery,
	according to tests/kernel/t_signal_and_fpu:

https://releng.netbsd.org/b5reports/alpha/2026/2026.07.21.22.53.03/test.html#kernel_t_signal_and_fpu_double

	The code appears to be there but it doesn't seem to be working:

https://nxr.netbsd.org/xref/src/sys/arch/alpha/alpha/machdep.c?r=1.385#1868

>How-To-Repeat:

	cd /usr/tests/kernel && atf-run t_signal_and_fpu

	(The test result in the float/double/ldouble tests is the
	number of iterations before the signal is delivered and the FPU
	registers appear to get trashed.)

>Fix:

	Yes, please!

>Release-Note:

>Audit-Trail:

Responsible-Changed-From-To: port-alpha-maintainer->thorpej
Responsible-Changed-By: thorpej@NetBSD.org
Responsible-Changed-When: Wed, 22 Jul 2026 22:00:00 +0000
Responsible-Changed-Why:
Take.


From: Taylor R Campbell <riastradh@NetBSD.org>
To: gnats-bugs@netbsd.org
Cc: thorpej@netbsd.org, port-alpha-maintainer@netbsd.org,
	netbsd-bugs@netbsd.org, gnats-admin@netbsd.org,
	thorpej@NetBSD.org
Subject: Re: port-alpha/60479 (alpha fails to save/restore fpu registers on signal delivery)
Date: Wed, 22 Jul 2026 22:12:53 +0000

 > Date: Wed, 22 Jul 2026 22:00:00 +0000 (UTC)
 > From: thorpej@NetBSD.org
 > 
 > Synopsis: alpha fails to save/restore fpu registers on signal delivery
 > 
 > Responsible-Changed-From-To: port-alpha-maintainer->thorpej
 > Responsible-Changed-By: thorpej@NetBSD.org
 > Responsible-Changed-When: Wed, 22 Jul 2026 22:00:00 +0000
 > Responsible-Changed-Why:
 > Take.

 Thanks!  Wonder if it's related to the /usr/tests/kernel/t_fpufork
 failure too.

From: Jason Thorpe <thorpej@me.com>
To: gnats-bugs@netbsd.org
Cc: Jason Thorpe <thorpej@netbsd.org>,
 gnats-admin@netbsd.org,
 netbsd-bugs@netbsd.org,
 campbell+netbsd@mumble.net
Subject: Re: port-alpha/60479 (alpha fails to save/restore fpu registers on
 signal delivery)
Date: Wed, 22 Jul 2026 20:05:16 -0700

 > On Jul 22, 2026, at 3:15=E2=80=AFPM, Taylor R Campbell via gnats =
 <gnats-admin@NetBSD.org> wrote:
 >>=20
 >> Synopsis: alpha fails to save/restore fpu registers on signal =
 delivery

 Taking a peek =E2=80=A6 =E2=80=9Care you sure?=E2=80=9D

 sendsig_signfo() does:

         cpu_getmcontext(l, &frame.sf_uc.uc_mcontext, =
 &frame.sf_uc.uc_flags);

 =E2=80=A6and cpu_getmcontext() does:

         /* Save floating point register context, if any, and copy it. */
         if (fpu_valid_p(l)) {
                 fpu_save(l);
                 (void)memcpy(&mcp->__fpregs, &pcb->pcb_fp,
                     sizeof (mcp->__fpregs));
                 mcp->__fpregs.__fp_fpcr =3D alpha_read_fp_c(l);
                 *flags |=3D _UC_FPU;
         }

 I guess I=E2=80=99ll have to fire one up and make sure ucontext / =
 mcontexts tests pass first.

 -- thorpej

From: Jason Thorpe <thorpej@me.com>
To: gnats-bugs@netbsd.org
Cc: port-alpha-maintainer@netbsd.org,
 gnats-admin@netbsd.org,
 netbsd-bugs@netbsd.org
Subject: Re: port-alpha/60479: alpha fails to save/restore fpu registers on
 signal delivery
Date: Fri, 24 Jul 2026 21:35:14 -0700

 > On Jul 22, 2026, at 1:55=E2=80=AFPM, campbell+netbsd@mumble.net via =
 gnats <gnats-admin@NetBSD.org> wrote:
 >=20
 >> Number:         60479
 >> Category:       port-alpha
 >> Synopsis:       alpha fails to save/restore fpu registers on signal =
 delivery
 >> Confidential:   no
 >> Severity:       serious
 >> Priority:       medium
 >> Responsible:    port-alpha-maintainer
 >> State:          open
 >> Class:          sw-bug
 >> Submitter-Id:   net
 >> Arrival-Date:   Wed Jul 22 20:55:00 +0000 2026
 >> Originator:     Taylor R Campbell
 >> Release:       =20
 >> Organization:
 > The AlphaBSD Fpundation, Inc.
 >> Environment:
 >> Description:
 >=20
 > FPU registers are not saved and restored on signal delivery,
 > according to tests/kernel/t_signal_and_fpu:
 >=20
 > =
 https://releng.netbsd.org/b5reports/alpha/2026/2026.07.21.22.53.03/test.ht=
 ml#kernel_t_signal_and_fpu_double

 I=E2=80=99m begnning to suspect this is a Qemu bug, because on real HW:

 [     1.000000] AlphaStation 200 4/233, 233MHz
 [     1.000000] 8192 byte page size, 1 processor.
 [     1.000000] total memory =3D 384 MB
 [     1.000000] (2000 KB reserved for PROM, 382 MB used by NetBSD)
 [     1.000000] avail memory =3D 362 MB
 [     1.000000] timecounter: Timecounters tick every 0.976 msec
 [     1.000000] Kernelized RAIDframe activated
 [     1.000000] mainbus0 (root)
 [     1.000000] cpu0 at mainbus0: ID 0 (primary), 21064A-0 (EV45)

 I get this:

 avanti:thorpej 16$ atf-run t_signal_and_fpu | atf-report
 Tests root: /usr/tests/kernel

 t_signal_and_fpu (1/1): 3 test cases
     double: [1.041472s] Passed.
     float: [1.040791s] Passed.
     ldouble: [1.041530s] Passed.
 [3.151730s]

 Summary for 1 test programs:
     3 passed test cases.
     0 failed test cases.
     0 expected failed test cases.
     0 skipped test cases.
 avanti:thorpej 17$=20

 -- thorpej

>Unformatted:

Home	PR Database Search