Patch for aic7xxx 2.4.0 test12 hang

Patch for aic7xxx 2.4.0 test12 hang

[Leslie Donaldson]

This is a temporary patch to keep the scsi driver from eating
your data.... I am working on a real fix....

Leslie Donaldson


*** linux/drivers/scsi/aic7xxx.c.2.4.0-12       Sat Jan  6 21:55:47 2001
--- linux/drivers/scsi/aic7xxx.c        Sat Jan  6 22:08:12 2001
***************
*** 7073,7078 ****
--- 7073,7092 ----
        else
        {
  
+         if(((strcmp("Adaptec AIC-7892 Ultra 160/m SCSI host
adapter",board_names[p->board_name_index])) == 0 ) ||
+            ((strcmp("Adaptec AIC-7899 Ultra 160/m SCSI host
adapter",board_names[p->board_name_index])) == 0 ))
+         {
+           /* The TCQ code for 160M devices is BROKEN */
+           /* This is a quick, dirty, sad fix until I have time for a
better */
+           /* one. We do it this way in case the driver is supporting
two seperate */
+           /* styles of scsi chipsets... I do have two boards so it
does happen */
+           tag_enabled = FALSE;
+           device->queue_depth = 3;  /* Tagged queueing is disabled. */
+           printk(INFO_LEAD "DISABLED TAGGED QUEUING, queue depth
%d.\n",
+                  p->host_no, device->channel, device->id,
+                  device->lun, 0);
+         }
+         else
          if (aic7xxx_tag_info[p->instance].tag_commands[tindex] == 255)
          {
            tag_enabled = FALSE;

-- 
/----------------------------\ Current Contractor: None
|    Leslie F. Donaldson     | Current Customer  : None
|    Computer Contractor     | Skills:
Unix/OS9/VMS/Linux/SUN-OS/C/C++/assembly
| Have Computer will travel. | WWW  :
http://www.cs.rose-hulman.edu/~donaldlf
\----------------------------/ Email: mail://donaldlf@cs.rose-hulman.edu
Goth Code V1.1: GoCS$$ TYg(T6,T9) B11Bk!^1 C6b-- P0(1,7) M+ a24 n---
b++:+
                H6'11" g m---- w+ r+++ D--~!% h+ s10 k+++ R-- Ssw
LusCA++
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Patch for aic7xxx 2.4.0 test12 hang

[Justin T. Gibbs]

>This is a temporary patch to keep the scsi driver from eating
>your data.... I am working on a real fix....
>
>Leslie Donaldson

What is the firmware revision of your Seagate drives?  There
were several models shipped in the recent past with firware that
would cause the drive to drop off the bus under high load.

There is also a known issue with U160 modes and the currently
embedded aic7xxx driver.  You might want to try the Adaptec
supported driver from here:

http://people.FreeBSD.org/~gibbs/linux/

6.09 BETA should be released later today.

--
Justin
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Patch for aic7xxx 2.4.0 test12 hang

[Leslie Donaldson]

"Justin T. Gibbs" wrote:
> 
> >This is a temporary patch to keep the scsi driver from eating
> >your data.... I am working on a real fix....
> >
> >Leslie Donaldson
> 
> What is the firmware revision of your Seagate drives? 

Not using seagate drives this is with quantum

see bottom....

> There
> were several models shipped in the recent past with firware that
> would cause the drive to drop off the bus under high load.

Ummm o the controller card...

00:06.0 SCSI storage controller: Adaptec 7899A (rev 01)
        Subsystem: Adaptec: Unknown device f620
        Flags: bus master, 66Mhz, medium devsel, latency 32, IRQ 16
        BIST result: 00
        I/O ports at 8000 [size=256]
        Memory at 000000000c864000 (64-bit, non-prefetchable) [size=4K]
        Expansion ROM at 000000000c800000 [disabled] [size=128K]
        Capabilities: [dc] Power Management version 2

00:06.1 SCSI storage controller: Adaptec 7899A (rev 01)
        Subsystem: Adaptec: Unknown device f620
        Flags: bus master, 66Mhz, medium devsel, latency 32, IRQ 23
        BIST result: 00
        I/O ports at 8400 [size=256]
        Memory at 000000000c865000 (64-bit, non-prefetchable) [size=4K]
        Expansion ROM at 000000000c820000 [disabled] [size=128K]
        Capabilities: [dc] Power Management version 2


> There is also a known issue with U160 modes and the currently
> embedded aic7xxx driver.  

That's true the problem is the TCQ command seems to be sequencing wrong.

>You might want to try the Adaptec
> supported driver from here:
> 
> http://people.FreeBSD.org/~gibbs/linux/
> 
> 6.09 BETA should be released later today.

I'll try it later  tonight, 

> --
> Justin


Jan 19 11:23:40 shadowdragon kernel: SCSI subsystem driver Revision:
1.00 
Jan 19 11:23:40 shadowdragon kernel: (scsi0) <Adaptec AIC-7899 Ultra
160/m SCSI host adapter> found at PCI 0/6/0 
Jan 19 11:23:40 shadowdragon kernel: (scsi0) Wide Channel A, SCSI ID=7,
32/255 SCBs 
Jan 19 11:23:40 shadowdragon kernel: (scsi0) Downloading sequencer
code... 392 instructions downloaded 
Jan 19 11:23:41 shadowdragon pcmcia: Starting PCMCIA services:
Jan 19 11:23:40 shadowdragon kernel: (scsi1) <Adaptec AIC-7899 Ultra
160/m SCSI host adapter> found at PCI 0/6/1 
Jan 19 11:23:40 shadowdragon kernel: (scsi1) Wide Channel B, SCSI ID=7,
32/255 SCBs 
Jan 19 11:23:40 shadowdragon kernel: (scsi1) Downloading sequencer
code... 392 instructions downloaded 
Jan 19 11:23:40 shadowdragon kernel: scsi0 : Adaptec AHA274x/284x/294x
(EISA/VLB/PCI-Fast SCSI) 5.2.1/5.2.0 
Jan 19 11:23:40 shadowdragon kernel:        <Adaptec AIC-7899 Ultra
160/m SCSI host adapter> 
Jan 19 11:23:40 shadowdragon kernel: scsi1 : Adaptec AHA274x/284x/294x
(EISA/VLB/PCI-Fast SCSI) 5.2.1/5.2.0 
Jan 19 11:23:40 shadowdragon kernel:        <Adaptec AIC-7899 Ultra
160/m SCSI host adapter> 
Jan 19 11:23:40 shadowdragon kernel: (scsi0:0:3:0) Synchronous at 5.0
Mbyte/sec, offset 8. 
Jan 19 11:23:40 shadowdragon kernel:   Vendor: FUJITSU   Model:
M2512A            Rev: 1408 
Jan 19 11:23:40 shadowdragon kernel:   Type:   Optical
Device                     ANSI SCSI revision: 02 
Jan 19 11:23:40 shadowdragon kernel: (scsi0:0:4:0) Synchronous at 10.0
Mbyte/sec, offset 8. 
Jan 19 11:23:40 shadowdragon kernel:   Vendor: TOSHIBA   Model: CD-ROM
XM-5701TA  Rev: 0167 
Jan 19 11:23:40 shadowdragon kernel:   Type:  
CD-ROM                             ANSI SCSI revision: 02 
Jan 19 11:23:40 shadowdragon kernel: (scsi1:0:0:0) Synchronous at 80.0
Mbyte/sec, offset 31. 
Jan 19 11:23:40 shadowdragon kernel:   Vendor: QUANTUM   Model: ATLAS
10K 18WLS   Rev: UC81 
Jan 19 11:23:40 shadowdragon kernel:   Type:  
Direct-Access                      ANSI SCSI revision: 03 
Jan 19 11:23:40 shadowdragon kernel: (scsi1:0:0:0) DISABLED TAGGED
QUEUING, queue depth 0. 
Jan 19 11:23:40 shadowdragon kernel: Detected scsi removable disk sda at
scsi0, channel 0, id 3, lun 0 
Jan 19 11:23:40 shadowdragon kernel: Detected scsi disk sdb at scsi1,
channel 0, id 0, lun 0 
Jan 19 11:23:40 shadowdragon kernel: sda : READ CAPACITY failed. 
Jan 19 11:23:40 shadowdragon kernel: sda : status = 1, message = 00,
host = 0, driver = 28  
Jan 19 11:23:40 shadowdragon kernel: sda : extended sense code = 2  
Jan 19 11:23:40 shadowdragon kernel: sda : block size assumed to be 512
bytes, disk size 1GB.   
Jan 19 11:23:40 shadowdragon kernel: Partition check: 
Jan 19 11:23:40 shadowdragon kernel:  sda: I/O error: dev 08:00, sector
0 
Jan 19 11:23:41 shadowdragon pcmcia:  modules
Jan 19 11:23:40 shadowdragon kernel:  unable to read partition table 
Jan 19 11:23:40 shadowdragon kernel: SCSI device sdb: 35877972 512-byte
hdwr sectors (18370 MB) 
Jan 19 11:23:40 shadowdragon kernel:  sdb: sdb1 sdb2 sdb3 sdb4 
Jan 19 11:23:40 shadowdragon kernel: Detected scsi CD-ROM sr0 at scsi0,
channel 0, id 4, lun 0 
Jan 19 11:23:40 shadowdragon kernel: Uniform CD-ROM driver Revision:
3.11 


-- 
/----------------------------\ Current Contractor: None
|    Leslie F. Donaldson     | Current Customer  : None
|    Computer Contractor     | Skills:
Unix/OS9/VMS/Linux/SUN-OS/C/C++/assembly
| Have Computer will travel. | WWW  :
http://www.cs.rose-hulman.edu/~donaldlf
\----------------------------/ Email: mail://donaldlf@cs.rose-hulman.edu
Goth Code V1.1: GoCS$$ TYg(T6,T9) B11Bk!^1 C6b-- P0(1,7) M+ a24 n---
b++:+
                H6'11" g m---- w+ r+++ D--~!% h+ s10 k+++ R-- Ssw
LusCA++
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Patch for aic7xxx 2.4.0 test12 hang

[Jeff Hartmann]

>> There is also a known issue with U160 modes and the currently
>> embedded aic7xxx driver.  
> 
> 
> That's true the problem is the TCQ command seems to be sequencing wrong.
> 
> 
>> You might want to try the Adaptec
>> supported driver from here:
>> 
>> http://people.FreeBSD.org/~gibbs/linux/
>> 
>> 6.09 BETA should be released later today.

Just a little FYI, I wanted to point out that 6.08 BETA fixed a problem 
I've been having since the 2.4.0-test series on a machine with the 
following adaptec integrated controller:
   Bus  4, device   7, function  0:
     SCSI storage controller: Adaptec 7899P (rev 1).
       IRQ 19.
       Master Capable.  Latency=64.  Min Gnt=40.Max Lat=25.
       I/O at 0x5000 [0x50ff].
       Non-prefetchable 64 bit memory at 0xf7e00000 [0xf7e00fff].
   Bus  4, device   7, function  1:
     SCSI storage controller: Adaptec 7899P (#2) (rev 1).
       IRQ 16.
       Master Capable.  Latency=64.  Min Gnt=40.Max Lat=25.
       I/O at 0x5400 [0x54ff].
       Non-prefetchable 64 bit memory at 0xf7f00000 [0xf7f00fff].

This is an Ultra160 controller I believe (or at least thats what it says 
during bootup.)

Before I applied this patch it would print garbage for the 
Vendor/Rev/Type/ANSI SCSI revision of my hard disk.  With this patch it 
does not.

I unfortunately know very little about SCSI drivers, so I can't say 
exactly what causes this problem with the stock 2.4.0 adaptec driver.

-Jeff

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Patch for aic7xxx 2.4.0 test12 hang

[Leslie Donaldson]

Jeff Hartmann wrote:
> 
> >> There is also a known issue with U160 modes and the currently
> >> embedded aic7xxx driver.
> >
> >
> > That's true the problem is the TCQ command seems to be sequencing wrong.
> >
> >
> >> You might want to try the Adaptec
> >> supported driver from here:
> >>
> >> http://people.FreeBSD.org/~gibbs/linux/
> >>
> >> 6.09 BETA should be released later today.
> 
> Just a little FYI, I wanted to point out that 6.08 BETA fixed a problem
> I've been having since the 2.4.0-test series on a machine with the
> following adaptec integrated controller:
>    Bus  4, device   7, function  0:
>      SCSI storage controller: Adaptec 7899P (rev 1).
>        IRQ 19.
>        Master Capable.  Latency=64.  Min Gnt=40.Max Lat=25.
>        I/O at 0x5000 [0x50ff].
>        Non-prefetchable 64 bit memory at 0xf7e00000 [0xf7e00fff].
>    Bus  4, device   7, function  1:
>      SCSI storage controller: Adaptec 7899P (#2) (rev 1).
>        IRQ 16.
>        Master Capable.  Latency=64.  Min Gnt=40.Max Lat=25.
>        I/O at 0x5400 [0x54ff].
>        Non-prefetchable 64 bit memory at 0xf7f00000 [0xf7f00fff].
> 
> This is an Ultra160 controller I believe (or at least thats what it says
> during bootup.)
> 
> Before I applied this patch it would print garbage for the
> Vendor/Rev/Type/ANSI SCSI revision of my hard disk.  With this patch it
> does not.
> 
> I unfortunately know very little about SCSI drivers, so I can't say
> exactly what causes this problem with the stock 2.4.0 adaptec driver.
> 
> -Jeff

  This sounds like the segate bios problem that Justin Gibbs refered to.
I have
two Western Digital enteprise drives that had the same problem. The
problem
that I was fighting was a Tagged Queue Command (or at least that is how
it is manifesting. This only occurs under heavy disk load for me.

My system is a 667 Alpha with a 64 bit PCI bus Feeding the 64 bit
Adaptec card.
(I note this cause I only know of one intel board with a 64 bit pci bus
and
 dual cpus' (found it on Tom's site back in November from the Tapei ??
computer
  conference.)

Well the problem seems to be that a command is flying and another is
then issued
causeing and overflow of the buffer causing the kernel to issue a
recurssive failure.

Error was

(scsi:1:0:0:0) Data overrun detected in Data-Out phase tag 5;
Have seen Data Phase. Length=0, Num SGS=0
Unable to handle kernel paging request at virtual address
003ffc0000006000
bzip2(8065): Oops 1

Then some NULL pointer stuff in the kernel.

Well the drive worked with no problem under the adaptec 3960UW
controller
so I finally caught the oops and played some with the TCQ logic.

The process that caused this was a rpm --rebuild of XFree. Something
which I have been doing lately as I am trying to get 4.0.2 to work
correctly. sigh.

I configured my patch to disable the TCQ logic only on 160M controllers
after other reports
of this problem. So far all the feed back says this fixed their problems
so that is were
I focused my attention.

Unfortuently emacs is segfaulting left and right these days so it has
slowed down my work a lot.
(rawhide systems do have a few ummm unstabilitys.)

I will try out the latest code sonnn and see if it crashes or works for
we and let everyone know.
My primary concern was to try to get a patch into 2.4.1 that at least
stops the kernel crashes.
Dosen't fix the problem but stops people from having an unstable
machine.

Well back to downloading.

Leslie Donaldson.

P.S. Thanks for all the input. It has been very helpful.


-- 
/----------------------------\ Current Contractor: None
|    Leslie F. Donaldson     | Current Customer  : None
|    Computer Contractor     | Skills:
Unix/OS9/VMS/Linux/SUN-OS/C/C++/assembly
| Have Computer will travel. | WWW  :
http://www.cs.rose-hulman.edu/~donaldlf
\----------------------------/ Email: mail://donaldlf@cs.rose-hulman.edu
Goth Code V1.1: GoCS$$ TYg(T6,T9) B11Bk!^1 C6b-- P0(1,7) M+ a24 n---
b++:+
                H6'11" g m---- w+ r+++ D--~!% h+ s10 k+++ R-- Ssw
LusCA++
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Patch for aic7xxx 2.4.0 test12 hang

[Leslie Donaldson]

Jeff Hartmann wrote:
> 
> >> There is also a known issue with U160 modes and the currently
> >> embedded aic7xxx driver.
> >
> >
> > That's true the problem is the TCQ command seems to be sequencing wrong.
> >
> >
> >> You might want to try the Adaptec
> >> supported driver from here:
> >>
> >> http://people.FreeBSD.org/~gibbs/linux/
> >>
> >> 6.09 BETA should be released later today.

> -Jeff

Hello all,
  I included everyone (boy + dog) so everyone knows the result.

 After much testing here is the results. I used the following patch

 linux-aic7xxx-6.0.9BETA-2.4.0.diffs

  against the ne 2.4.0 kernel sans my patch and happy to report it
cleaned
up the TCQ problem. I beat on my scsi system fairly hard.
3 makes of X with -j 10 going. and a a few ls -lR / running
in the background with the periodic sync command.

Know the question left is when will this be part of the kernel. If it is
going
to be a few versions I would like to suggest my patch goes into the
current 2.4.0
and 2.5.0 trees. If this version is fixing to be sent in then forget
about it.
My primary concern is right now the 160M controllers are unstable and
with the new
distributions fixing to roll this is going to cause people to be very
ummm mad.

I leave it up to the powers that be.

Leslie Donaldson

-- 
/----------------------------\ Current Contractor: None
|    Leslie F. Donaldson     | Current Customer  : None
|    Computer Contractor     | Skills:
Unix/OS9/VMS/Linux/SUN-OS/C/C++/assembly
| Have Computer will travel. | WWW  :
http://www.cs.rose-hulman.edu/~donaldlf
\----------------------------/ Email: mail://donaldlf@cs.rose-hulman.edu
Goth Code V1.1: GoCS$$ TYg(T6,T9) B11Bk!^1 C6b-- P0(1,7) M+ a24 n---
b++:+
                H6'11" g m---- w+ r+++ D--~!% h+ s10 k+++ R-- Ssw
LusCA++
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Patch for aic7xxx 2.4.0 test12 hang

[Gregory McLean]

On Sun, 21 Jan 2001, Leslie Donaldson wrote:

> Jeff Hartmann wrote:
> > 
> > >> There is also a known issue with U160 modes and the currently
> > >> embedded aic7xxx driver.
> > >
> > >
> > > That's true the problem is the TCQ command seems to be sequencing wrong.
> > >
> > >
> > >> You might want to try the Adaptec
> > >> supported driver from here:
> > >>
> > >> http://people.FreeBSD.org/~gibbs/linux/
> > >>
> > >> 6.09 BETA should be released later today.

I'll give the 6.09 driver a try 6.08 locks up on me after some time.


Gregory McLean     | "Why do they call it a TV set when you only get one?"
globix.net         | - George Carlin 
Core Engineer      |
Network Services   |
PGP Key: 0xE94D1363|
---------------This is my opinion, All MINE! I tell you-----------------

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

Re: Patch for aic7xxx 2.4.0 test12 hang

[Justin T. Gibbs]

>Hello all,
>  I included everyone (boy + dog) so everyone knows the result.
>
> After much testing here is the results. I used the following patch
>
> linux-aic7xxx-6.0.9BETA-2.4.0.diffs
>
> against the ne 2.4.0 kernel sans my patch and happy to report it
> cleaned up the TCQ problem.

Thanks for the positive report!

I can't say exactly what kernel release will have the new driver
in it, but work is well underway toward making that happen.

--
Justin
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/