Date: Tue, 1 Oct 2002 09:12:20 -0400 (EDT) From: Andrew Gallatin <gallatin@cs.duke.edu> To: Peter Grehan <grehan@freebsd.org> Cc: Andrew Gallatin <gallatin@cs.duke.edu>, freebsd-ppc@freebsd.org Subject: Re: atomic.h vs atomic.s Message-ID: <15769.40756.959761.373890@grasshopper.cs.duke.edu> In-Reply-To: <3D999C61.DCA16E15@freebsd.org> References: <15768.30116.381592.891890@grasshopper.cs.duke.edu> <15768.62718.937391.963035@grasshopper.cs.duke.edu> <3D98FD84.D523B8F@freebsd.org> <15769.39791.486131.152431@grasshopper.cs.duke.edu> <3D999C61.DCA16E15@freebsd.org>
index | next in thread | previous in thread | raw e-mail
Peter Grehan writes:
> > Are you saying that the lwarx instruction itself is not atomic wrt to
> > interrupts?
>
> No, I got it wrong, see my previous email.
>
> > More or less.. I could compile about 20 files, then the machine would
> > crash. Once I made it through all the .o's, I couldn't get a kernel
> > to link without the machine falling over, so I gave up and used the
> > cross compiler/linker that Mark Peek posted last year..
>
> OK, more bugs to fix :-)
No, I think it was the same one, but the other direction. When
linking a kernel, I ran out of mbufs. Probably because the ref
counts never dropped to zero on some clusters, so they never got
freed. The crashes every 10-15 files were panic: m_ext refcnt < 0.
# rm kernel.debug
# make kernel.debug
linking kernel.debug
Out of mbuf address space!
Consider increasing NMBCLUSTERS
exclusive sleep mutex Giant r = 0 (0x33b680) locked @../../../vm/vm_object.c:446
panic: mb_reclaim() called with locks held
syncing disks... 393 All mbufs or mbuf clusters exhausted, please see tuning(7).
<...>
BTW, ddb works well enough to be useful just by fixing the compile
errors (booting w/o loader) and commenting out a NetBSD (db_symstr)
function I didn't want to think about:
panic: m_ext refcnt < 0
Welcome to Debugger, panic
Stopped at 0x2b1da4: lwz r0, r1, 0x14
db> tr
0xe74c1be0: at panic+0x134
0xe74c1ca0: at m_freem+0x104
0xe74c1cf0: at gem_reset+0x2400
0xe74c1d30: at gem_intr+0xe0
0xe74c1d60: at swi_sched+0x340
0xe74c1da0: at fork_exit+0x100
0xe74c1dd0: at fork_trampoline+0xc
saved LR(0x7c0802a2) is invalid.
db> e 0xe093c284
-0x1f6c3d7c: 0
Index: db_disasm.c
===================================================================
RCS file: /home/ncvs/src/sys/powerpc/powerpc/db_disasm.c,v
retrieving revision 1.1
diff -u -r1.1 db_disasm.c
--- db_disasm.c 10 Jul 2002 12:21:54 -0000 1.1
+++ db_disasm.c 1 Oct 2002 00:01:15 -0000
@@ -717,8 +717,8 @@
LI = LI << 8;
LI = LI >> 6;
LI += loc;
- db_symstr(pstr, LI, DB_STGY_ANY);
- pstr += strlen(pstr);
+/* db_symstr(pstr, LI, DB_STGY_ANY);
+ pstr += strlen(pstr);*/
func &= ~Op_LI;
}
switch (func & Op_SIMM) {
Index: trap.c
===================================================================
RCS file: /home/ncvs/src/sys/powerpc/powerpc/trap.c,v
retrieving revision 1.33
diff -u -r1.33 trap.c
--- trap.c 19 Sep 2002 04:38:35 -0000 1.33
+++ trap.c 30 Sep 2002 23:35:32 -0000
@@ -280,7 +280,7 @@
printtrap(frame->exc, frame, 1, (frame->srr1 & PSL_PR));
#ifdef DDB
- if ((debugger_on_panic || db_active) && kdb_trap(frame->exc, 0, frame))
+ if (kdb_trap(frame->exc, frame))
return;
#endif
panic("%s trap", trapname(frame->exc));
The thing which really scares me is the transient abort traps which go
away when the compiler is re-run:
# make ffs_softdep.o
cc -c -O -pipe -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual -fformat-extensions -ansi -g -nostdinc -I- -I. -I../../.. -I../../../dev -I../../../contrib/dev/acpica -I../../../contrib/ipfilter -D_KERNEL -include opt_global.h -msoft-float -fno-common -ffreestanding -Wno-error ../../../ufs/ffs/ffs_softdep.c
cc: Internal error: Abort trap (program as)
Please submit a full bug report.
See <URL:http://www.gnu.org/software/gcc/bugs.html>; for instructions.
*** Error code 1
Stop in /usr/src/sys/powerpc/compile/G4.
# make ffs_softdep.o
cc -c -O -pipe -Wall -Wredundant-decls -Wnested-externs -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith -Winline -Wcast-qual -fformat-extensions -ansi -g -nostdinc -I- -I. -I../../.. -I../../../dev -I../../../contrib/dev/acpica -I../../../contrib/ipfilter -D_KERNEL -include opt_global.h -msoft-float -fno-common -ffreestanding -Wno-error ../../../ufs/ffs/ffs_softdep.c
#
Perhaps this is another consequence of the atomic problem... I'll try
your kernel and see if the problems go away.
Drew
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-ppc" in the body of the message
help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?15769.40756.959761.373890>