NetBSD Problem Report #59294

From www@netbsd.org  Sun Apr 13 23:37:17 2025
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) server-digest SHA256
	 client-signature RSA-PSS (2048 bits) client-digest SHA256)
	(Client CN "mail.NetBSD.org", Issuer "mail.NetBSD.org CA" (not verified))
	by mollari.NetBSD.org (Postfix) with ESMTPS id B38C51A9239
	for <gnats-bugs@gnats.NetBSD.org>; Sun, 13 Apr 2025 23:37:17 +0000 (UTC)
Message-Id: <20250413233716.315081A923D@mollari.NetBSD.org>
Date: Sun, 13 Apr 2025 23:37:16 +0000 (UTC)
From: campbell+netbsd@mumble.net
Reply-To: campbell+netbsd@mumble.net
To: gnats-bugs@NetBSD.org
Subject: tests/lib/libc/gen/t_siginfo: sigfpe_flt raises SIGILL, not SIGFPE
X-Send-Pr-Version: www-1.0

>Number:         59294
>Category:       port-hppa
>Synopsis:       tests/lib/libc/gen/t_siginfo: sigfpe_flt raises SIGILL, not SIGFPE
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    port-hppa-maintainer
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Apr 13 23:40:00 +0000 2025
>Last-Modified:  Mon Apr 14 00:50:02 +0000 2025
>Originator:     Taylor R Campbell
>Release:        current
>Organization:
The HP PA-BSD Sigfpeinformation
>Environment:
>Description:
Relevant excerpt from atf-run output:

tp-start: 1744585134.484021, t_siginfo, 1
tc-start: 1744585134.522808, sigfpe_flt
tc-se:Test program crashed; attempting to get stack trace
tc-se:[New process 492]
tc-se:Core was generated by `t_siginfo'.
tc-se:Program terminated with signal SIGILL, Illegal instruction.
tc-se:#0  0x000119d0 in atfu_sigfpe_flt_body (tc=<optimized out>) at /home/riastradh/netbsd/current/src/tests/lib/libc/gen/t_siginfo.c:334
tc-se:#0  0x000119d0 in atfu_sigfpe_flt_body (tc=<optimized out>) at /home/riastradh/netbsd/current/src/tests/lib/libc/gen/t_siginfo.c:334
tc-se:#1  0xaf80bc7c in atf_tc_run (tc=<optimized out>, resfile=<optimized out>) at /usr/src/external/bsd/atf/dist/atf-c/tc.c:1024
tc-se:#2  0xaf80745c in atf_tp_run (tp=<optimized out>, tcname=<optimized out>, resfile=0xafed8020 "/tmp/atf-run.ArQ50d/tcr") at /usr/src/external/bsd/atf/dist/atf-c/tp.c:205
tc-se:#3  0xaf806d08 in run_tc (exitcode=<synthetic pointer>, p=0xb00017c8, tp=0xb00017f8) at /usr/src/external/bsd/atf/dist/atf-c/detail/tp_main.c:510
tc-se:#4  controlled_main (exitcode=<synthetic pointer>, add_tcs_hook=@0x14172: 0x11d1c <atfu_tp_add_tcs>, argv=<optimized out>, argc=<optimized out>) at /usr/src/external/bsd/atf/dist/atf-c/detail/tp_main.c:580
tc-se:#5  atf_tp_main (argc=<optimized out>, argv=<optimized out>, add_tcs_hook=@0x14172: 0x11d1c <atfu_tp_add_tcs>) at /usr/src/external/bsd/atf/dist/atf-c/detail/tp_main.c:610
tc-se:#6  0x000124a0 in main (argc=<optimized out>, argv=<optimized out>) at /home/riastradh/netbsd/current/src/tests/lib/libc/gen/t_siginfo.c:542
tc-se:Stack trace complete
tc-end: 1744585175.995480, sigfpe_flt, failed, Test program received signal 4 (core dumped)
tp-end: 1744585176.47816, t_siginfo

Signal 4 is SIGILL: https://nxr.NetBSD.org/xref/src/sys/sys/signal.h?r=1.78#55

gdb on the core dump from `/usr/tests/lib/libc/gen/t_siginfo sigfpe_flt' confirms that it is in fact the fdiv,dbl instruction at issue:

Program terminated with signal SIGILL, Illegal instruction.
#0  0x000119d0 in atfu_sigfpe_flt_body (tc=<optimized out>)
    at /home/riastradh/netbsd/current/src/tests/lib/libc/gen/t_siginfo.c:334
334     /home/riastradh/netbsd/current/src/tests/lib/libc/gen/t_siginfo.c: No such file or directory.
(gdb) x/i $pc
=> 0x119d0 <atfu_sigfpe_flt_body+312>:  fdiv,dbl fr7,fr22,fr7

>How-To-Repeat:
cd /usr/tests/lib/libc/gen
atf-run t_siginfo | atf-report
>Fix:
Yes, please!

>Audit-Trail:
From: Taylor R Campbell <riastradh@NetBSD.org>
To: gnats-bugs@NetBSD.org, netbsd-bugs@NetBSD.org
Cc: 
Subject: Re: port-hppa/59294: tests/lib/libc/gen/t_siginfo: sigfpe_flt raises SIGILL, not SIGFPE
Date: Mon, 14 Apr 2025 00:46:46 +0000

 Might be another qemu hppa fpu bug.  I printed the exception registers
 (32-bit halves of fr0-fr3) in the T_EXCEPTION | T_USER case of trap in
 hppa/trap.c, and:

 [  52.9324218] fpp[0]=0x0
 [  52.9324218] fpp[1]=0x0
 [  52.9450154] fpp[2]=0x0
 [  52.9450154] fpp[3]=0x0
 [  52.9450154] fpp[4]=0x0
 [  52.9450154] fpp[5]=0x0
 [  52.9450154] fpp[6]=0x0
 [  52.9450154] fpp[7]=0x0
 [  52.9450154] trap:829: type=0x4e ex=0x0 inst=0x0

 I don't think they're supposed to be all zero!  I think they're
 supposed to have copies of the instructions that triggered the
 exception.  (Except for the first one, which is the fp status
 register.)

Home	PR Database Search