aic7xxx freezes with kernel 2.4.13

aic7xxx freezes with kernel 2.4.13

[Pim Zandbergen]

Hi,

I've got a Dell PowerEdge 1300 with dual PIII's and dual aic7xxx
controllers. One controller is onboard, the other is in a PCI slot.

The system is running Red Hat 7.1 with kernel 2.4.13.

Lately, this system is experiencing freezes that may last one or two
minutes. These usually occur during heavy Samba activity. After the
freeze, the system usually recovers, but by then, the Samba clients
have timed out their operations.

Syslog shows the freezes are related to the SCSI subsystem. I'm having
trouble interpreting this information. Is my hardware suspect or could
this be a driver bug?

Syslog entries (with aic7xxx=verbose) showing the boot process and a
system freeze can be found on

http://www.macroscoop.nl/~pim/aic7xxx/syslog.html (98.080 bytes) or
http://www.macroscoop.nl/~pim/aic7xxx/syslog.gz   (6.410 bytes)

Thanks,
Pim

Re: aic7xxx freezes with kernel 2.4.13

[Cristian CONSTANTIN]

On Wed, Nov 07, 2001 at 04:20:17PM +0100, Pim Zandbergen wrote:
> Hi,
> 
> I've got a Dell PowerEdge 1300 with dual PIII's and dual aic7xxx
> controllers. One controller is onboard, the other is in a PCI slot.
> 
> The system is running Red Hat 7.1 with kernel 2.4.13.
> 
> Lately, this system is experiencing freezes that may last one or two
> minutes. These usually occur during heavy Samba activity. After the
> freeze, the system usually recovers, but by then, the Samba clients
> have timed out their operations.
> 
> Syslog shows the freezes are related to the SCSI subsystem. I'm having
> trouble interpreting this information. Is my hardware suspect or could
> this be a driver bug?
> 
> Syslog entries (with aic7xxx=verbose) showing the boot process and a
> system freeze can be found on
> 
> http://www.macroscoop.nl/~pim/aic7xxx/syslog.html (98.080 bytes) or
> http://www.macroscoop.nl/~pim/aic7xxx/syslog.gz   (6.410 bytes)
> 

cristian: same problem here. posted some time ago on linux-kernel and
some days ago on linux-scsi. no solution so far...

my config, in a very few words:
- tyan SMP motherboard + 2xAthlon
- adaptec aic7xxx (2 onboard 1 PCI - tried with all of them, same
  result)
- IBM HDD - mines are the 36 GB model - 
http://www.storage.ibm.com/hdd/ultra/ul36lzx.htm 
it seems yours are the 18 GB model - 
http://www.storage.ibm.com/hdd/ultra/ul18es.htm  
(more details in the postings on linux-kernel, linux-scsi)

is adaptec+ibm a "dangerous" combination??
what is strange is that the "freezing" behaviour appears randomly, even
though it seems to be connected to "heavy" accesses to the HDD.

please let me know if you have found a solution for this problem,

thanks!
bye now!
-- 
 _          |(4) Any person who knowingly and without authorization uses,
(_'_        |accesses or attempts to access any computer, computer system,
  (_'rist   |computer network, or any computer software, program,
            |documentation or data contained in such computer, computer
            |system or computer network, commits computer crime.
            |                --Oregon Computer Crime Law
GPG public key at https://www.mobile-ip.de/~crist/constantin.key

Re: aic7xxx freezes with kernel 2.4.13

[Justin T. Gibbs]

>On Wed, Nov 07, 2001 at 04:20:17PM +0100, Pim Zandbergen wrote:
>> Hi,
>> 
>> I've got a Dell PowerEdge 1300 with dual PIII's and dual aic7xxx
>> controllers. One controller is onboard, the other is in a PCI slot.
>> 
>> The system is running Red Hat 7.1 with kernel 2.4.13.
>> 
>> Lately, this system is experiencing freezes that may last one or two
>> minutes. These usually occur during heavy Samba activity. After the
>> freeze, the system usually recovers, but by then, the Samba clients
>> have timed out their operations.
>> 
>> Syslog shows the freezes are related to the SCSI subsystem. I'm having
>> trouble interpreting this information. Is my hardware suspect or could
>> this be a driver bug?
>> 
>> Syslog entries (with aic7xxx=verbose) showing the boot process and a
>> system freeze can be found on
>> 
>> http://www.macroscoop.nl/~pim/aic7xxx/syslog.html (98.080 bytes) or
>> http://www.macroscoop.nl/~pim/aic7xxx/syslog.gz   (6.410 bytes)

In the case of these traces, it appears that the bus is not up to
snuff for the negotiated speed.  We're in data-out phase, our DMA
engine is enabled, our FIFO is full of data to send, but the target
has not requested any more data.  This means that either the controller
missed a REQ or the target missed an ACK.  I would verify your cabling
and termination.

>cristian: same problem here. posted some time ago on linux-kernel and
>some days ago on linux-scsi. no solution so far...
>
>my config, in a very few words:
>- tyan SMP motherboard + 2xAthlon
>- adaptec aic7xxx (2 onboard 1 PCI - tried with all of them, same
>  result)

I'd need to see the logs to know for sure that what you are seeing
is the same as Cristian.

--
Justin

Re: aic7xxx freezes with kernel 2.4.13 - solved

[Pim Zandbergen]

Hi,

For the archives: my problem is solved. The problem went away after
removing a (crashed) IDE disk and the HPT370 IDE controller it was
connected to. I replaced the faulty IDE disk with a SCSI disk and
now all my IDE and SCSI problems are gone.

Thanks for your input,
Pim

> >On Wed, Nov 07, 2001 at 04:20:17PM +0100, Pim Zandbergen wrote:
> >> Hi,
> >> 
> >> I've got a Dell PowerEdge 1300 with dual PIII's and dual aic7xxx
> >> controllers. One controller is onboard, the other is in a PCI slot.
> >> 
> >> The system is running Red Hat 7.1 with kernel 2.4.13.
> >> 
> >> Lately, this system is experiencing freezes that may last one or two
> >> minutes. These usually occur during heavy Samba activity. After the
> >> freeze, the system usually recovers, but by then, the Samba clients
> >> have timed out their operations.
> >> 
> >> Syslog shows the freezes are related to the SCSI subsystem. I'm having
> >> trouble interpreting this information. Is my hardware suspect or could
> >> this be a driver bug?
> >> 
> >> Syslog entries (with aic7xxx=verbose) showing the boot process and a
> >> system freeze can be found on
> >> 
> >> http://www.macroscoop.nl/~pim/aic7xxx/syslog.html (98.080 bytes) or
> >> http://www.macroscoop.nl/~pim/aic7xxx/syslog.gz   (6.410 bytes)
> 
> In the case of these traces, it appears that the bus is not up to
> snuff for the negotiated speed.  We're in data-out phase, our DMA
> engine is enabled, our FIFO is full of data to send, but the target
> has not requested any more data.  This means that either the controller
> missed a REQ or the target missed an ACK.  I would verify your cabling
> and termination.