2.6.3-mm1 and aic7xxx

2.6.3-mm1 and aic7xxx

[Fabio Coatti]

I'm experiencing some problems with 2.6.3-mm1 release: on boot the system 
hangs trying to detect scsi devices, and the light on scsi cdrom flashes 
every few seconds. With 2.6.3-rc3-mm1 all works just fine, and I get this 
syslog entry:

Feb 19 22:23:15 kefk kernel: ahc_pci:3:6:0: Host Adapter Bios disabled.  Using 
default SCSI device parameters
Feb 19 22:23:15 kefk kernel: scsi0 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA 
DRIVER, Rev 6.2.36
Feb 19 22:23:15 kefk kernel:         <Adaptec 2902/04/10/15/20C/30C SCSI 
adapter>
Feb 19 22:23:15 kefk kernel:         aic7850: Single Channel A, SCSI Id=7, 
3/253 SCBs

<<<<<<<<<<<<<<<2.6.3-mm1 hangs here

Feb 19 22:23:15 kefk kernel:
Feb 19 22:23:15 kefk kernel:   Vendor: Nikon     Model: COOLSCANIII       Rev: 
1.31
Feb 19 22:23:15 kefk kernel:   Type:   Scanner                            ANSI 
SCSI revision: 02
Feb 19 22:23:15 kefk kernel: (scsi0:A:3): 10.000MB/s transfers (10.000MHz, 
offset 15)
Feb 19 22:23:15 kefk kernel:   Vendor: PLEXTOR   Model: CD-ROM PX-40TS    Rev: 
1.01
Feb 19 22:23:15 kefk kernel:   Type:   CD-ROM                             ANSI 
SCSI revision: 02
Feb 19 22:23:15 kefk kernel: (scsi0:A:5): 10.000MB/s transfers (10.000MHz, 
offset 15)
Feb 19 22:23:15 kefk kernel:   Vendor: YAMAHA    Model: CRW6416S          Rev: 
1.0c
Feb 19 22:23:15 kefk kernel:   Type:   CD-ROM                             ANSI 
SCSI revision: 02

I've also noticed (only with 2.6.3-mm1) a "PCI BIOS passed non existent PCI 
BUS 0!" message when it probes ICH5, i.e.

Feb 19 22:23:15 kefk kernel: ICH5: IDE controller at PCI slot 0000:00:1f.1

<<<<<<<<<<<<HERE

Feb 19 22:23:15 kefk kernel: ICH5: chipset revision 2
Feb 19 22:23:15 kefk kernel: ICH5: not 100%% native mode: will probe irqs 
later
Feb 19 22:23:15 kefk kernel:     ide0: BM-DMA at 0xf000-0xf007, BIOS settings: 
hda:DMA, hdb:pio
Feb 19 22:23:15 kefk kernel:     ide1: BM-DMA at 0xf008-0xf00f, BIOS settings: 
hdc:pio, hdd:pio

system: PIV 2.HT/i875p (abit ic7-g), SMP kernel (SMT), 
gcc (GCC) 3.3.1 (Mandrake Linux 9.2 3.3.1-2mdk)

I can provide more details if needed, and any help will be appreciated.

Thanks.

-- 
Fabio Coatti       http://www.ferrara.linux.it/members/cova     
Ferrara Linux Users Group           http://ferrara.linux.it
GnuPG fp:9765 A5B6 6843 17BC A646  BE8C FA56 373A 5374 C703
Old SysOps never die... they simply forget their password.

Re: 2.6.3-mm1 and aic7xxx

[Andrew Morton]

Fabio Coatti <cova@ferrara.linux.it> wrote:
>
> I'm experiencing some problems with 2.6.3-mm1 release: on boot the system 
> hangs trying to detect scsi devices, and the light on scsi cdrom flashes 
> every few seconds. With 2.6.3-rc3-mm1 all works just fine, and I get this 
> syslog entry:
> 
> Feb 19 22:23:15 kefk kernel: ahc_pci:3:6:0: Host Adapter Bios disabled.  Using 
> default SCSI device parameters
> Feb 19 22:23:15 kefk kernel: scsi0 : Adaptec AIC7XXX EISA/VLB/PCI SCSI HBA 
> DRIVER, Rev 6.2.36
> Feb 19 22:23:15 kefk kernel:         <Adaptec 2902/04/10/15/20C/30C SCSI 
> adapter>
> Feb 19 22:23:15 kefk kernel:         aic7850: Single Channel A, SCSI Id=7, 
> 3/253 SCBs
> 
> <<<<<<<<<<<<<<<2.6.3-mm1 hangs here

Are you able to get a sysrq-T or sysrq-P trace?

> I've also noticed (only with 2.6.3-mm1) a "PCI BIOS passed non existent PCI 
> BUS 0!" message when it probes ICH5, i.e.

Could be an acpi thing.  If you have time, could you try

	patch -p1 -R < bk-acpi.patch

and see if that helps?

ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.3/2.6.3-mm1/broken-out/bk-acpi.patch

Thanks.

Re: 2.6.3-mm1 and aic7xxx

[Fabio Coatti]

Alle 01:21, venerdì 20 febbraio 2004, Andrew Morton ha scritto:

> > Feb 19 22:23:15 kefk kernel:         aic7850: Single Channel A, SCSI
> > Id=7, 3/253 SCBs
> >
> > <<<<<<<<<<<<<<<2.6.3-mm1 hangs here
>
> Are you able to get a sysrq-T or sysrq-P trace?

I've just tried, but the system seems completely hung. The only activity is 
from scsi cdrom device, that blinks a light every 20/30 seconds, more or 
less. Even Caps/Num/Scroll lock keyboards led are frozen.

(tried also with mm2)
>
> > I've also noticed (only with 2.6.3-mm1) a "PCI BIOS passed non existent
> > PCI BUS 0!" message when it probes ICH5, i.e.
>
> Could be an acpi thing.  If you have time, could you try
>
> 	patch -p1 -R < bk-acpi.patch
>
> and see if that helps?

At this moment I can't reach kernel.org, some connectivity issue over the 
ocean, I suppose :)
I'll try again in few hours.

-- 
Fabio Coatti       http://www.ferrara.linux.it/members/cova     
Ferrara Linux Users Group           http://ferrara.linux.it
GnuPG fp:9765 A5B6 6843 17BC A646  BE8C FA56 373A 5374 C703
Old SysOps never die... they simply forget their password.

Re: 2.6.3-mm1 and aic7xxx

[Fabio Coatti]

Alle 01:21, venerdì 20 febbraio 2004, Andrew Morton ha scritto:

> > I've also noticed (only with 2.6.3-mm1) a "PCI BIOS passed non existent
> > PCI BUS 0!" message when it probes ICH5, i.e.
>
> Could be an acpi thing.  If you have time, could you try
>
> 	patch -p1 -R < bk-acpi.patch
>
> and see if that helps?

Tried, the error message is disappeared (but my kernel still hangs on scsi 
detection, so I'm unable if this has other effects :) )


-- 
Fabio Coatti       http://www.ferrara.linux.it/members/cova     
Ferrara Linux Users Group           http://ferrara.linux.it
GnuPG fp:9765 A5B6 6843 17BC A646  BE8C FA56 373A 5374 C703
Old SysOps never die... they simply forget their password.

Re: 2.6.3-mm1 and aic7xxx

[P. Christeas]

Just noted some nasty behavior at 2.6.3 ( not -mm1) wrt. aic7xxx.

With 2.6.2 I hadn't noticed any trouble (for the last 6 days I've been running 
it). Now, the box hung twice, while trying to use the scsi cd-recorder (HP 
m820e) I have. The machine is 'blind' (i.e. no console) so I couldn't get any 
trace/logs..
First time was while starting 'cdrecord-2.0'. Just when trying to query the 
device. At the same time the scsi bus was accessing an iomega zip100 drive, 
reading data.
Second time must have been the device scanning done by k3b (I believe it uses 
the same scsi libs as cdrecord). Note also that I have an atapi (not 
ide-scsi) recorder, which had been fixating the disk it had written when the 
box hung.

Does this relate to the problem you mention ? 

 I could only recover the following log: (0 4 0 is the scsi M820e recorder)

SCSI error : <0 0 4 0> return code = 0x70002
Feb 21 12:07:19 pfn1 last message repeated 6 times
scsi0:0:4:0: Attempting to queue an ABORT message
CDB: 0x46 0x0 0x0 0x0 0x24 0x0 0x0 0x0 0x8 0x0
scsi0: At time of recovery, card was not paused
>>>>>>>>>>>>>>>>>> Dump Card State Begins <<<<<<<<<<<<<<<<<
scsi0: Dumping Card State in Command phase, at SEQADDR 0x166
Card was paused
ACCUM = 0x80, SINDEX = 0xac, DINDEX = 0xc0, ARG_2 = 0x0
HCNT = 0x0 SCBPTR = 0x0
SCSISIGI[0x84]:(BSYI|CDI) ERROR[0x0] SCSIBUSL[0x8] 
LASTPHASE[0x80]:(CDI) SCSISEQ[0x12]:(ENAUTOATNP|ENRSELI) 
SBLKCTL[0x0] SCSIRATE[0x28] SEQCTL[0x10]:(FASTMODE) 
SEQ_FLAGS[0x0] SSTAT0[0x7]:(DMADONE|SPIORDY|SDONE) 
SSTAT1[0x2]:(PHASECHG) SSTAT2[0x0] SSTAT3[0x0] SIMODE0[0x0] 
SIMODE1[0xac]:(ENSCSIPERR|ENBUSFREE|ENSCSIRST|ENSELTIMO) 
SXFRCTL0[0x88]:(SPIOEN|DFON) DFCNTRL[0x4]:(DIRECTION) 
DFSTATUS[0x6d]:(FIFOEMP|DFTHRESH|HDONE|FIFOQWDEMP|DFCACHETH) 
STACK: 0x35 0xd3 0x160 0x1a1
SCB count = 4
Kernel NEXTQSCB = 2
Card NEXTQSCB = 2
QINFIFO entries: 
Waiting Queue entries: 
Disconnected Queue entries: 
QOUTFIFO entries: 
Sequencer Free SCB List: 1 2 
Sequencer SCB Info: 
  0 SCB_CONTROL[0x0] SCB_SCSIID[0x47] SCB_LUN[0x0] SCB_TAG[0x3] 
  1 SCB_CONTROL[0x0] SCB_SCSIID[0xff]:(TWIN_CHNLB|OID|TWIN_TID) 
SCB_LUN[0xff]:(SCB_XFERLEN_ODD|LID) SCB_TAG[0xff] 
  2 SCB_CONTROL[0x0] SCB_SCSIID[0xff]:(TWIN_CHNLB|OID|TWIN_TID) 
SCB_LUN[0xff]:(SCB_XFERLEN_ODD|LID) SCB_TAG[0xff] 
Pending list: 
  3 SCB_CONTROL[0x0] SCB_SCSIID[0x47] SCB_LUN[0x0] 
Kernel Free SCB list: 1 0 
Untagged Q(4): 3 
DevQ(0:4:0): 0 waiting
DevQ(0:6:0): 0 waiting

<<<<<<<<<<<<<<<<< Dump Card State Ends >>>>>>>>>>>>>>>>>>
scsi0:0:4:0: Device is active, asserting ATN
Recovery code sleeping
Recovery code awake
Timer Expired
aic7xxx_abort returns 0x2003
scsi0:0:4:0: Attempting to queue a TARGET RESET message
CDB: 0x46 0x0 0x0 0x0 0x24 0x0 0x0 0x0 0x8 0x0
aic7xxx_dev_reset returns 0x2003
Recovery SCB completes
... hang, I press the win98 key .. 
Feb 21 12:09:54 pfn1 syslogd 1.4.1: restart.

Re: 2.6.3-mm1 and aic7xxx

[Len Brown]

On Fri, 2004-02-20 at 19:53, Fabio Coatti wrote:
> Alle 01:21, venerdì 20 febbraio 2004, Andrew Morton ha scritto:
> 
> > > I've also noticed (only with 2.6.3-mm1) a "PCI BIOS passed non
> existent
> > > PCI BUS 0!" message when it probes ICH5, i.e.
> >
> > Could be an acpi thing.  If you have time, could you try
> >
> >       patch -p1 -R < bk-acpi.patch
> >
> > and see if that helps?
> 
> Tried, the error message is disappeared (but my kernel still hangs on
> scsi 
> detection, so I'm unable if this has other effects :) )

Fabio,
Any chance you can isolate further where this broke by finding the
latest release where it worked properly?

ie. does vanilla 2.6.3 work if you back out the mm patch?

If 2.6.3 works, then I'd be interested if the following 2.6.3 patch
breaks it:
http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/2.6.3/acpi-20040116-2.6.3.diff.gz

If 2.6.3 fails, does 2.6.2 work?
If 2.6.2 works, I'd be interested if either of the following break it:
http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/2.6.2/acpi-20040211-2.6.2.diff.bz2
http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/2.6.2/acpi-20040116-2.6.2.diff.gz

This will tell us if the most recent ACPI changes are causing this
failure, or if it is something else.

thanks,
-Len

ps, note that you can recover the original source tree by using patch -R
on the patch that you applied.

Re: 2.6.3-mm1 and aic7xxx

[Fabio Coatti]

Alle Monday 23 February 2004 08:28, Len Brown ha scritto:
>
> Fabio,
> Any chance you can isolate further where this broke by finding the
> latest release where it worked properly?
>
> ie. does vanilla 2.6.3 work if you back out the mm patch?
>
> If 2.6.3 works, then I'd be interested if the following 2.6.3 patch
> breaks it:

[SNIP]

Sure, this evening I'll try to catch the exact last release that worked fine. 
I've used only  mm releases so far, so I've noticed the problem only with 
2.6.3-mm1 and mm2; in a few hours I'll report back for other versions as you 
suggested.


-- 
Fabio Coatti       http://www.ferrara.linux.it/members/cova     
Ferrara Linux Users Group           http://ferrara.linux.it
GnuPG fp:9765 A5B6 6843 17BC A646  BE8C FA56 373A 5374 C703
Old SysOps never die... they simply forget their password.

Re: 2.6.3-mm1 and aic7xxx

[Fabio Coatti]

Alle 08:28, lunedì 23 febbraio 2004, Len Brown ha scritto:

>
> Fabio,
> Any chance you can isolate further where this broke by finding the
> latest release where it worked properly?
>
> ie. does vanilla 2.6.3 work if you back out the mm patch?

I don't know if this come late (I've been quite busy) anyway I've found that 
vanilla 2.6.3 and 2.6.2 works just fine; mmX versions hangs (mm1, mm2, etc..)
The latest working mm version is 2.6.3-rc3-mm1
I've also tried 2.6.4-rc1 and 2.6.4-rc1-mm1: 2.6.4-rc1 boots and works, -mm1 
hang at the very same point.

> If 2.6.3 works, then I'd be interested if the following 2.6.3 patch
> breaks it:
>http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/2.6.3/acpi-20040116-2.6.3.diff.gz

Sorry, but I can't find this patch, maybe it's outdated; if you can give me a 
new link, I'll try asap.

HTH

-- 
Fabio Coatti       http://www.ferrara.linux.it/members/cova     
Ferrara Linux Users Group           http://ferrara.linux.it
GnuPG fp:9765 A5B6 6843 17BC A646  BE8C FA56 373A 5374 C703
Old SysOps never die... they simply forget their password.

Re: 2.6.3-mm1 and aic7xxx

[Andrew Morton]

Fabio Coatti <cova@ferrara.linux.it> wrote:
>
> 
>  >
>  > Fabio,
>  > Any chance you can isolate further where this broke by finding the
>  > latest release where it worked properly?
>  >
>  > ie. does vanilla 2.6.3 work if you back out the mm patch?
> 
>  I don't know if this come late (I've been quite busy) anyway I've found that 
>  vanilla 2.6.3 and 2.6.2 works just fine; mmX versions hangs (mm1, mm2, etc..)
>  The latest working mm version is 2.6.3-rc3-mm1
>  I've also tried 2.6.4-rc1 and 2.6.4-rc1-mm1: 2.6.4-rc1 boots and works, -mm1 
>  hang at the very same point.
> 
>  > If 2.6.3 works, then I'd be interested if the following 2.6.3 patch
>  > breaks it:
>  >http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/2.6.3/acpi-20040116-2.6.3.diff.gz
> 
>  Sorry, but I can't find this patch, maybe it's outdated; if you can give me a 
>  new link, I'll try asap.

If you have the time, please test 2.6.4-rc1 and then test 2.6.4-rc1
plus

ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.4-rc1/2.6.4-rc1-mm1/broken-out/bk-acpi.patch

thanks.

RE: 2.6.3-mm1 and aic7xxx

[Brown, Len]

Fabio,
I think this failure is from something else in -mm besides ACPI.
Otherwise, 2.6.4-rc1 wouldn't be working for you -- because 2.6.4
includes the ACPI patch you were running when you first hit this.

>I've also tried 2.6.4-rc1 and 2.6.4-rc1-mm1: 2.6.4-rc1 boots 
>and works, -mm1 hang at the very same point.

To verify with the latest software, please apply the latest ACPI patch
to 2.6.4-rc1:

http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/
2.6.4/acpi-20040220-2.6.4.diff.gz

If it works, then something else in -mm is causing your problem.  If it
fails, then something in the latest ACPI patch (which is included in
-mm1) is causing the failure.

Thanks,
-Len

Re: 2.6.3-mm1 and aic7xxx

[Fabio Coatti]

Alle 00:42, lunedì 1 marzo 2004, Andrew Morton ha scritto:

>
> If you have the time, please test 2.6.4-rc1 and then test 2.6.4-rc1
> plus
>
> ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.4-rc1/2.6
>.4-rc1-mm1/broken-out/bk-acpi.patch

Done:
2.6.4-rc1 works, and the same holds for 2.6.4-rc1 with bk-acpi.patch applied.
2.6.4-rc1-mm1 hangs.


-- 
Fabio Coatti       http://www.ferrara.linux.it/members/cova     
Ferrara Linux Users Group           http://ferrara.linux.it
GnuPG fp:9765 A5B6 6843 17BC A646  BE8C FA56 373A 5374 C703
Old SysOps never die... they simply forget their password.

Re: 2.6.3-mm1 and aic7xxx

[Fabio Coatti]

Alle 00:45, lunedì 1 marzo 2004, Brown, Len ha scritto:

> To verify with the latest software, please apply the latest ACPI patch
> to 2.6.4-rc1:
>
> http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/
> 2.6.4/acpi-20040220-2.6.4.diff.gz
>
> If it works, then something else in -mm is causing your problem.  If it
> fails, then something in the latest ACPI patch (which is included in
> -mm1) is causing the failure.

Tried, and with the patch all works just fine. So it can be something else, 
applied in -mm tree after 2.6.3-rc3-mm1, tha causes the failure; i'll try to 
have a look at changes, trying to find which one causes the problem.


-- 
Fabio Coatti       http://www.ferrara.linux.it/members/cova     
Ferrara Linux Users Group           http://ferrara.linux.it
GnuPG fp:9765 A5B6 6843 17BC A646  BE8C FA56 373A 5374 C703
Old SysOps never die... they simply forget their password.

Re: 2.6.3-mm1 and aic7xxx

[Fabio Coatti]

Alle 01:39, lunedì 1 marzo 2004, Fabio Coatti ha scritto:
> Alle 00:45, lunedì 1 marzo 2004, Brown, Len ha scritto:
> > To verify with the latest software, please apply the latest ACPI patch
> > to 2.6.4-rc1:
> >
> > http://ftp.kernel.org/pub/linux/kernel/people/lenb/acpi/patches/release/
> > 2.6.4/acpi-20040220-2.6.4.diff.gz
> >
> > If it works, then something else in -mm is causing your problem.  If it
> > fails, then something in the latest ACPI patch (which is included in
> > -mm1) is causing the failure.
>
> Tried, and with the patch all works just fine. So it can be something else,
> applied in -mm tree after 2.6.3-rc3-mm1, tha causes the failure; i'll try
> to have a look at changes, trying to find which one causes the problem.

I've just tried, as suggested by john stultz <johnstul@us.ibm.com> in  thread 
"2.6.3-mm3 hangs on  boot x440 (scsi?)":

>> Index: arch/i386/kernel/acpi/boot.c
>> ===================================================================
>> RCS file: /var/cvs/linux-2.6/arch/i386/kernel/acpi/boot.c,v
>> retrieving revision 1.10
>> diff -u -p -r1.10 boot.c
>> --- a/arch/i386/kernel/acpi/boot.c    17 Feb 2004 12:51:46 -0000      1.10
>> +++ b/arch/i386/kernel/acpi/boot.c    26 Feb 2004 16:34:12 -0000
>> @@ -506,24 +461,17 @@ acpi_boot_init (void)
>>
>>       acpi_lapic = 1;
>>
>> -#endif /*CONFIG_X86_LOCAL_APIC*/
>> +#endif /* CONFIG_X86_LOCAL_APIC */
>>
>>  #if defined(CONFIG_X86_IO_APIC) && defined(CONFIG_ACPI_INTERPRETER)
>>
>>       /*
>>        * I/O APIC
>> -      * --------
>>        */
>>
>> -     /*
>> -      * ACPI interpreter is required to complete interrupt setup,
>> -      * so if it is off, don't enumerate the io-apics with ACPI.
>> -      * If MPS is present, it will handle them,
>> -      * otherwise the system will stay in PIC mode
>> -      */
>> -     if (acpi_disabled || acpi_noirq) {
>> +     if (acpi_noirq) {
>>               return 1;
>> -        }
>> +     }
>>
>>       /*
>>        * if "noapic" boot option, don't look for IO-APICs
>
>
>That chunk shouldn't drop the "if (acpi_disabled ..." bit.
>Adding that check back in fixes it for me.

In fact, this fixes 2.6.4-rc1-mm1 and now I'm writing using that kernel. I 
don't knoe w if this is a quick hack or a definitive solution, but now mmX 
series works again for me, and scsi devices seems to be correctly handled.




-- 
Fabio Coatti       http://www.ferrara.linux.it/members/cova     
Ferrara Linux Users Group           http://ferrara.linux.it
GnuPG fp:9765 A5B6 6843 17BC A646  BE8C FA56 373A 5374 C703
Old SysOps never die... they simply forget their password.