Some potential FPU emulator problem

Some potential FPU emulator problem

[Thiemo Seufer]

Hello All,

I used the appended patch which enables the assert() in the
FPU emulator. The result (running in qemu) is:

[snip]
Will now check root file system:fsck 1.40-WIP (14-Nov-2006)
[/sbin/fsck.ext3 (1) -- /] fsck.ext3 -a -C0 /dev/hda1
/dev/hda1: clean, 21846/131072 files, 371295/524272 blocks
.
EXT3 FS on hda1, internal journal
Setting the system clock..
System Clock set. Local time: Tue May  1 22:43:28 UTC 2007.
Cleaning up ifupdown....
Loading device-mapper support.
Will now check all file systems.
fsck 1.40-WIP (14-Nov-2006)
Checking all file systems.
Done checking file systems.
A log is being saved in /var/log/fsck/checkfs if that location is writable.
Setting kernel variables...done.
Will now mount local filesystems:.
Will now activate swapfile swap:done.
BUG: soft lockup detected on CPU#0!
Call Trace:
[<80016d98>] dump_stack+0x8/0x34
[<800594ac>] softlockup_tick+0x114/0x178
[<800405bc>] update_process_times+0x38/0x90
[<80016510>] timer_interrupt+0x120/0x150
[<80059b60>] handle_IRQ_event+0x6c/0xdc
[<8005b650>] handle_level_irq+0xd4/0x17c
[<80010b70>] plat_irq_dispatch+0xb0/0x150
[<80011200>] ret_from_irq+0x0/0x4
[<80249a28>] _spin_unlock_irqrestore+0x28/0x48
[<80042ed0>] force_sig_info+0xa4/0xf8
[<80017f1c>] do_tr+0xb4/0x134
[<800111e0>] ret_from_exception+0x0/0x20
[<80021dac>] ieee754sp_format+0x28/0x50c
[<80027e14>] ieee754sp_fdp+0x94/0x3d0
[<80020194>] fpu_emulator_cop1Handler+0x1334/0x1b40
[<80017934>] do_cpu+0x344/0x380
[<800111e0>] ret_from_exception+0x0/0x20

Without this patch the same configuration boot fine. I don't know i
f this just means one of the asserts is broken or if there is
something wrong with the FPU emulation.


Thiemo


diff --git a/arch/mips/math-emu/ieee754dp.h b/arch/mips/math-emu/ieee754dp.h
index a37370d..f777806 100644
--- a/arch/mips/math-emu/ieee754dp.h
+++ b/arch/mips/math-emu/ieee754dp.h
@@ -28,7 +28,7 @@
 
 #include "ieee754int.h"
 
-#define assert(expr) ((void)0)
+#define assert(expr) BUG_ON(expr)
 
 /* 3bit extended double precision sticky right shift */
 #define XDPSRS(v,rs)	\
diff --git a/arch/mips/math-emu/ieee754sp.h b/arch/mips/math-emu/ieee754sp.h
index ae82f51..68bd4c0 100644
--- a/arch/mips/math-emu/ieee754sp.h
+++ b/arch/mips/math-emu/ieee754sp.h
@@ -28,7 +28,7 @@
 
 #include "ieee754int.h"
 
-#define assert(expr) ((void)0)
+#define assert(expr) BUG_ON(expr)
 
 /* 3bit extended single precision sticky right shift */
 #define SPXSRSXn(rs) \

Re: Some potential FPU emulator problem

[Atsushi Nemoto]

On Tue, 1 May 2007 23:52:14 +0100, Thiemo Seufer <ths@networkno.de> wrote:
> [<80042ed0>] force_sig_info+0xa4/0xf8
> [<80017f1c>] do_tr+0xb4/0x134
> [<800111e0>] ret_from_exception+0x0/0x20
> [<80021dac>] ieee754sp_format+0x28/0x50c
> [<80027e14>] ieee754sp_fdp+0x94/0x3d0
> [<80020194>] fpu_emulator_cop1Handler+0x1334/0x1b40
> [<80017934>] do_cpu+0x344/0x380
> [<800111e0>] ret_from_exception+0x0/0x20

Is this kernel contain commit ba755f8ec80fdbf2b5212622eabf7355464c6327 ?
I suppose the fix might show you a little bit better result.

> Without this patch the same configuration boot fine. I don't know i
> f this just means one of the asserts is broken or if there is
> something wrong with the FPU emulation.

Perhaps the triggered assert() is the first one in ieee754sp_format(),
but I do not know if it was a proper test...

---
Atsushi Nemoto