Re: still no luck with Radeon or Adaptec SCSI controller in 2.4.X

Re: still no luck with Radeon or Adaptec SCSI controller in 2.4.X

[Kevin B. Hendricks]

Hi,

> You have to have the new driver off or as a module.

Aha!  Thanks, I should have thought of that.  I recompiled with the old
AIC-7XXX driver and now SCSI works like a charm!  No more wierd aborts.

The new driver and my setup just don't get along.

So with that fixed, I tried to get the network working but ran into
trouble.

- I missed loading the ipchains module since I was running ipchains on
2.2.19

- I missed the fact that my two ethernet cards are now recognized in
reverse order (eth1 is the internal BMAC+ when it was eth0 previously).

This of-course messed up all my ipchains and masquerading.

Arrghhh!  Anyway it is working now!

Thanks,

Kevin

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: still no luck with Radeon or Adaptec SCSI controller in 2.4.X

[Tom Rini]

On Sun, Oct 07, 2001 at 09:44:13PM -0400, Kevin B. Hendricks wrote:
> Hi,
>
> > You have to have the new driver off or as a module.
>
> Aha!  Thanks, I should have thought of that.  I recompiled with the old
> AIC-7XXX driver and now SCSI works like a charm!  No more wierd aborts.
>
> The new driver and my setup just don't get along.
>
> So with that fixed, I tried to get the network working but ran into
> trouble.
>
> - I missed loading the ipchains module since I was running ipchains on
> 2.2.19

IP Tables, and then ipchains compat. :)

> - I missed the fact that my two ethernet cards are now recognized in
> reverse order (eth1 is the internal BMAC+ when it was eth0 previously).

This is what modules are for. :)
/etc/modules.conf:
alias eth0 bmac
alias eth1 tulip  # or de4x5, since tulip is broken on some cards

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: still no luck with Radeon or Adaptec SCSI controller in 2.4.X

[Tom Rini]

On Sun, Oct 07, 2001 at 10:00:43PM -0400, Kevin B. Hendricks wrote:
> Hi Tom,
>
> > This is what modules are for. :)
> > /etc/modules.conf:
> > alias eth0 bmac
> > alias eth1 tulip  # or de4x5, since tulip is broken on some cards
>
> Ah...I never thought of that.
>
> By the way, do you know of any way to open a shell session in one specific
> framebuffer?

If you have 2 you mean?  video=map:000111 will put tty[1-3] on fb0 and
tty[4-6] on fb1.

--
Tom Rini (TR1265)
http://gate.crashing.org/~trini/

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

broken startup for scsi in 2.4.12 rsync'd this morning

[Kevin B. Hendricks]

Hi Ben,

I just worked through things with Justin on how to work around an issue in
the aic7xxx driver and all worked with your 2.4.11 rsynced yesterday
sometime.

Today (this morning) I rsynced and got your 2.4.12 kernel.

I can not longer get past the initialization of the adaptec card (with or
without Justin's extra kernel parameters).  After it accesses the adaptec
card, it simply does a kernel panic.

I compared dmesg output and both at using Justin's 6.2.1 version of the
aic7xxx driver.

Do you know of any other changes from your 2.4.12 kernel from your 2.4.11
kernel that might impact use of an adaptec pic scsi card?

I did try rebuilding after an "make mrproper" just to make sure and the
same problems happened.  I have no trouble booting your 2.4.11 kernel with
Justin's aic7xxx kernel options.

Any ideas here on how I track this one down would be greatly appreciated.

Thanks,

Kevin


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: broken startup for scsi in 2.4.12 rsync'd this morning

[Kevin B. Hendricks]

Hi,

In case this helps.  Here is the error message:

scsi0:A:1: ahc_intr-referenced scb not valid during seqint 0x71 scb(0)

This was copied by hand before the kernel panic timeout of 180 seconds was
done.

Anyone with any ideas?

Thanks,

Kevin


On October 13, 2001 01:42, Kevin B. Hendricks wrote:
> Hi Ben,
>
> I just worked through things with Justin on how to work around an issue
> in the aic7xxx driver and all worked with your 2.4.11 rsynced yesterday
> sometime.
>
> Today (this morning) I rsynced and got your 2.4.12 kernel.
>
> I can not longer get past the initialization of the adaptec card (with
> or without Justin's extra kernel parameters).  After it accesses the
> adaptec card, it simply does a kernel panic.
>
> I compared dmesg output and both at using Justin's 6.2.1 version of the
> aic7xxx driver.
>
> Do you know of any other changes from your 2.4.12 kernel from your
> 2.4.11 kernel that might impact use of an adaptec pic scsi card?
>
> I did try rebuilding after an "make mrproper" just to make sure and the
> same problems happened.  I have no trouble booting your 2.4.11 kernel
> with Justin's aic7xxx kernel options.
>
> Any ideas here on how I track this one down would be greatly
> appreciated.
>
> Thanks,
>
> Kevin
>
>

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: broken startup for scsi in 2.4.12 rsync'd this morning

[Benjamin Herrenschmidt]

>I just worked through things with Justin on how to work around an issue in
>the aic7xxx driver and all worked with your 2.4.11 rsynced yesterday
>sometime.
>
>Today (this morning) I rsynced and got your 2.4.12 kernel.
>
>I can not longer get past the initialization of the adaptec card (with or
>without Justin's extra kernel parameters).  After it accesses the adaptec
>card, it simply does a kernel panic.
>
>I compared dmesg output and both at using Justin's 6.2.1 version of the
>aic7xxx driver.
>
>Do you know of any other changes from your 2.4.12 kernel from your 2.4.11
>kernel that might impact use of an adaptec pic scsi card?
>
>I did try rebuilding after an "make mrproper" just to make sure and the
>same problems happened.  I have no trouble booting your 2.4.11 kernel with
>Justin's aic7xxx kernel options.
>
>Any ideas here on how I track this one down would be greatly appreciated.

I've neem myself experiencing various kernel lockups at boot with SCSI
enabled & aic7xxx. In fact, I had some problems with .11 too with a
disk not reponding on my dual G4.

I don't know what's up, but I'd apreciate if you could send me the
panic infos so I can start tracing it.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: broken startup for scsi in 2.4.12 rsync'd this morning

[Kevin B. Hendricks]

Hi Ben,

> I've neem myself experiencing various kernel lockups at boot with SCSI
> enabled & aic7xxx. In fact, I had some problems with .11 too with a
> disk not reponding on my dual G4.
>
> I don't know what's up, but I'd apreciate if you could send me the
> panic infos so I can start tracing it.

Unforutnately there really isn't any kind of backtrace info given by the
panic.

All that it says is (hand copied)

scsi0:A:1 ahc_intr - referenced scb not vlaid during seqint 0x71 scb(0)

scsio: Dumping Card State in Message - in phase at SEQADDR 0x1c1

...

Kernel panic: for safety
In interrupt handler - not syncing
<0>Rebooting in 180 seconds.

I found the line in the aic7xxx.c  file which prints this.  It seems
ahc_lookup_scb() returns 0 and then it does the kernel panic.

I can't seem to find any backtrace info for the System map. and none of
the messages get written to log - I guess because no sync is done?

I would be happy to copy down all of the Adaptec debug info but I am not
sure it will help.

Is there a way to force ithe kernel to drop into some sort of debug mode
upon doing a "panic"?

Thanks,

Kevin

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

now random kernel panics with aic7xxx driver 6.2.4 in kernel 2.4.12

[Kevin B. Hendricks]

[-- Attachment #1: Type: text/plain, Size: 693 bytes --]

Hi Justin,

Well, the idea of passing tag_info to the aic7xxx did help me boot but now
I am getting random kernel panics.  So I upgraded to your latest 6.2.4
with Ben's 2.4.12 and overnight when no activity was present, I got a
panic.

I rebooted and tried again and got another panic.

I have attached the relevant information from  /var/log/messages.
It seems to involve a different drive now on that same card.

Would you please take a look at the log and let me know if there is
anything that jumps out at you?

If not, I will simply downgrade to the old driver since I really need a
stable system right now and then try your driver again later sometime in
the future.

Thanks.

Kevin





[-- Attachment #2: aic624.log.gz --]
[-- Type: application/x-gzip, Size: 2068 bytes --]

Re: now random kernel panics with aic7xxx driver 6.2.4 in kernel 2.4.12

[Justin T. Gibbs]

>Hi Justin,
>
>Well, the idea of passing tag_info to the aic7xxx did help me boot but now
>I am getting random kernel panics.  So I upgraded to your latest 6.2.4
>with Ben's 2.4.12 and overnight when no activity was present, I got a
>panic.

It would be interesting to know what type of panics you saw prior
to 6.2.4.  It might help me track down where this problem began.

>Would you please take a look at the log and let me know if there is
>anything that jumps out at you?

The card seems to be loosing some of its resources.  I can't say
why yet.  I think it may be a regression in some of the abort
code in the kernel.  Could you try running with AHC_DEBUG_SEQUENCER
enabled?  You can just #define this to 1 in aic7xxx.c.

--
Justin

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: now random kernel panics with aic7xxx driver 6.2.4 in kernel 2.4.12

[Kevin B. Hendricks]

Hi Justin,

> It would be interesting to know what type of panics you saw prior
> to 6.2.4.  It might help me track down where this problem began.

This is my first foray into 2.4.X kernels.  I did not have any panics with
the old aic7xxx driver.  I first used 6.2.1 in kernel 2.4.11.  So you
pretty much know the whole story.  The reason I moved to your 6.2.4 was
that at some point I could not even boot 2.4.12 since it came back with:

scsi0:A:1: ahc_intr - referenced scb not valid during seqint 0x71 scb(0)
...
Kernel panic: for safety
In interrupt handler - not syncing
<0>Rebooting in 180 seconds

But the message logs never caught any of the error message since a sync
was not done.

> The card seems to be loosing some of its resources.  I can't say
> why yet.  I think it may be a regression in some of the abort
> code in the kernel.  Could you try running with AHC_DEBUG_SEQUENCER
> enabled?  You can just #define this to 1 in aic7xxx.c.

I have to go out of town tonight but I will rebuild with this define and
send you a full error message log on Wedsnesday.

Thanks, I hope we can get this nailed down.  I really would rather use
your new driver.

Kevin


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: now random kernel panics with aic7xxx driver 6.2.4 in kernel 2.4.12

[Benjamin Herrenschmidt]

>Hi Justin,
>
>> It would be interesting to know what type of panics you saw prior
>> to 6.2.4.  It might help me track down where this problem began.
>
>This is my first foray into 2.4.X kernels.  I did not have any panics with
>the old aic7xxx driver.  I first used 6.2.1 in kernel 2.4.11.  So you
>pretty much know the whole story.  The reason I moved to your 6.2.4 was
>that at some point I could not even boot 2.4.12 since it came back with:
>
>scsi0:A:1: ahc_intr - referenced scb not valid during seqint 0x71 scb(0)
>...
>Kernel panic: for safety
>In interrupt handler - not syncing
><0>Rebooting in 180 seconds
>
>But the message logs never caught any of the error message since a sync
>was not done.

I've been getting bizqrre problems with aic7xxx since 2.4.11 on my
dual G4 as well. It started with some problems detecting my external
disk, but the problem disappeared by itself using a more recent
kernel (during the .11-pre series).

However when I first compiled .12, the machine locked up during boot
just after printing "SCSI subsystem driver Revision: 1.00". It was
not a hard lockup as the console cursor was still flashing, looks like
one of the CPUs got hung or something around those lines.

I recompiled this kernel with xmon (our small kernel debugger we have
on PPC), the problem went away magically...

Today, I recompiled that same kernel, but with different drivers built-in
(nothing related to SCSI, but that did change the kernel size and boot
timing), I'm getting a  similar error as above (referenced scb not valid)
followed by a panic (using the 6.2.1 driver).

I'll try to do the tests you suggest asap, I hope to have some time
tonight or tomorrow. Note that I used earlier 2.4.x kernels with your
driver on that same box (with the same 2 adaptec cards & disk setup)
without problems.

Ben.


** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: now random kernel panics with aic7xxx driver 6.2.4 in kernel 2.4.12

[Justin T. Gibbs]

>Today, I recompiled that same kernel, but with different drivers built-in
>(nothing related to SCSI, but that did change the kernel size and boot
>timing), I'm getting a  similar error as above (referenced scb not valid)
>followed by a panic (using the 6.2.1 driver).

Do you know if anything changed in the PCI code that might affect
the initialization of PCI devices?  It is almost as if some of our
PCI register writes are not being honored (e.g. prefetching has
suddenly been enabled for this card even though the prefetch bit
if off in all of its BARs).  6.2.1 has been in the kernel for some
time and was not changed in going from .10 -> .12.

--
Justin

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: now random kernel panics with aic7xxx driver 6.2.4 in kernel 2.4.12

[Kevin B. Hendricks]

Hi,

I can confirm this for my machine.  It seems recompiling and changing
unrelated things does seem to impact the performance of the driver.

Since changing the memory footprint seems to impact the driver, I was
wondering about a few things:

- Is hardware "DMA" being used to swap in and out scb blocks of memory?
If so, are there scb memory "alignment" issues at work here when doing DMA?

- what about issues with locking when doing the DMA to prevent other reads
and or writes colliding?  Ben's trouble with SMP machines might be
indicative of something here?

- data cache flushing issues?

Is there any specific code we should focus on in the driver when trying to
figure this thing out?

Thanks,

Kevin

On October 15, 2001 11:33, Justin T. Gibbs wrote:
> >Today, I recompiled that same kernel, but with different drivers
> > built-in (nothing related to SCSI, but that did change the kernel size
> > and boot timing), I'm getting a  similar error as above (referenced
> > scb not valid) followed by a panic (using the 6.2.1 driver).
>
> Do you know if anything changed in the PCI code that might affect
> the initialization of PCI devices?  It is almost as if some of our
> PCI register writes are not being honored (e.g. prefetching has
> suddenly been enabled for this card even though the prefetch bit
> if off in all of its BARs).  6.2.1 has been in the kernel for some
> time and was not changed in going from .10 -> .12.

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/

Re: now random kernel panics with aic7xxx driver 6.2.4 in kernel 2.4.12

[Justin T. Gibbs]

>Hi Justin,
>
>Well, the idea of passing tag_info to the aic7xxx did help me boot but now
>I am getting random kernel panics.  So I upgraded to your latest 6.2.4
>with Ben's 2.4.12 and overnight when no activity was present, I got a
>panic.

I haven't forgotten about this problem... just been dragged off onto
another fire for a few days.

Can you apply this patch and reproduce the problem again?  Hopefully
this additional diagnostic information will help me understand the
problem.

--
Justin

diff -c -r aic7xxx/aic7xxx.c /usr/src/sys/dev/aic7xxx/aic7xxx.c
*** aic7xxx/aic7xxx.c	Wed Oct 24 17:19:14 2001
--- /usr/src/sys/dev/aic7xxx/aic7xxx.c	Wed Oct 24 18:28:48 2001
***************
*** 6302,6308 ****
  	printf("ACCUM = 0x%x, SINDEX = 0x%x, DINDEX = 0x%x, ARG_2 = 0x%x\n",
  	       ahc_inb(ahc, ACCUM), ahc_inb(ahc, SINDEX), ahc_inb(ahc, DINDEX),
  	       ahc_inb(ahc, ARG_2));
! 	printf("HCNT = 0x%x\n", ahc_inb(ahc, HCNT));
  	printf("SCSISEQ = 0x%x, SBLKCTL = 0x%x\n",
  	       ahc_inb(ahc, SCSISEQ), ahc_inb(ahc, SBLKCTL));
  	printf(" DFCNTRL = 0x%x, DFSTATUS = 0x%x\n",
--- 6302,6309 ----
  	printf("ACCUM = 0x%x, SINDEX = 0x%x, DINDEX = 0x%x, ARG_2 = 0x%x\n",
  	       ahc_inb(ahc, ACCUM), ahc_inb(ahc, SINDEX), ahc_inb(ahc, DINDEX),
  	       ahc_inb(ahc, ARG_2));
! 	printf("HCNT = 0x%x SCBPTR = 0x%x\n", ahc_inb(ahc, HCNT),
! 	       ahc_inb(ahc, SCBPTR));
  	printf("SCSISEQ = 0x%x, SBLKCTL = 0x%x\n",
  	       ahc_inb(ahc, SCSISEQ), ahc_inb(ahc, SBLKCTL));
  	printf(" DFCNTRL = 0x%x, DFSTATUS = 0x%x\n",
***************
*** 6375,6380 ****
--- 6376,6392 ----
  	}
  	printf("\n");

+ 	printf("Sequencer SCB Info: ");
+ 	for (i = 0; i < ahc->scb_data->maxhscbs; i++) {
+ 		ahc_outb(ahc, SCBPTR, i);
+ 		printf("%d(c 0x%x, s 0x%x, l %d, t 0x%x) ",
+ 		       i, ahc_inb(ahc, SCB_CONTROL),
+ 		       ahc_inb(ahc, SCB_SCSIID),
+ 		       ahc_inb(ahc, SCB_LUN),
+ 		       ahc_inb(ahc, SCB_TAG));
+ 	}
+ 	printf("\n");
+
  	printf("Pending list: ");
  	i = 0;
  	LIST_FOREACH(scb, &ahc->pending_scbs, pending_links) {
***************
*** 6382,6388 ****
  			break;
  		if (scb != LIST_FIRST(&ahc->pending_scbs))
  			printf(", ");
! 		printf("%d", scb->hscb->tag);
  		if ((ahc->flags & AHC_PAGESCBS) == 0) {
  			ahc_outb(ahc, SCBPTR, scb->hscb->tag);
  			printf("(0x%x, 0x%x)", ahc_inb(ahc, SCB_CONTROL),
--- 6394,6401 ----
  			break;
  		if (scb != LIST_FIRST(&ahc->pending_scbs))
  			printf(", ");
! 		printf("%d(c 0x%x, s 0x%x, l %d)", scb->hscb->tag,
! 		       scb->hscb->control, scb->hscb->scsiid, scb->hscb->lun);
  		if ((ahc->flags & AHC_PAGESCBS) == 0) {
  			ahc_outb(ahc, SCBPTR, scb->hscb->tag);
  			printf("(0x%x, 0x%x)", ahc_inb(ahc, SCB_CONTROL),

** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/