Re: [PATCH] aacraid: Panics during init time reset (Was: [PATCH] aacraid: [Fastboot] Panics for AACRAID driver during 'insmod' for kexec test)

[Judith Lebzelter <judith@linux-foundation.org>]

On Mon, Apr 02, 2007 at 04:29:35PM -0400, Salyzyn, Mark wrote:
> Cool! Look forward to your results.
> 
> Sorry for being so snitty sounding with the 5 minute comment, I was just
> admiring the pace of activity!

No problem.  I thought it was funny.:)

Judith
> 
> Sincerely -- Mark Salyzyn
> 
> > -----Original Message-----
> > From: Judith Lebzelter [mailto:judith@linux-foundation.org] 
> > Sent: Monday, April 02, 2007 4:17 PM
> > To: Salyzyn, Mark
> > Cc: Judith Lebzelter; James Bottomley; 
> > linux-scsi@vger.kernel.org; Duane Cox
> > Subject: Re: [PATCH] aacraid: Panics during init time reset 
> > (Was: [PATCH] aacraid: [Fastboot] Panics for AACRAID driver 
> > during 'insmod' for kexec test)
> > 
> > 
> > On Mon, Apr 02, 2007 at 03:56:50PM -0400, Salyzyn, Mark wrote:
> > > Judith, another layer on this onion also discovered by Duane, the
> > > interrupt enable handler also needed to be set ... The 
> > interrupt enable
> > > was called from within the synchronous command handler.
> > > 
> > > My private email with the fix was sent a whole 5 minutes 
> > ahead of yours
> > > ;-> Here is the jist of it for the observers:
> > 
> > I saw it the second I hit 'submit'.
> > 
> > > 
> > >     /* Failure to reset here is an option ... */
> > >     dev->a_ops.adapter_sync_cmd = rx_sync_cmd;
> > > +   dev->a_ops.adapter_enable_int = aac_rx_disable_interrupt;
> > >     dev->OIMR = status = rx_readb (dev, MUnit.OIMR);
> > > 
> > > Yes, the disable interrupt method patched into the enable 
> > int platform
> > > function. Later init code will reset it accordingly.
> > > 
> > > Sincerely -- Mark Salyzyn
> > 
> > Yes, that works now.  I will run a series to check if the 
> > other IRQ interrupt 
> > problem is resolved.
> > 
> > Judith
> > 
> > > 
> > > 
> > > > -----Original Message-----
> > > > From: Judith Lebzelter [mailto:judith@linux-foundation.org] 
> > > > Sent: Monday, April 02, 2007 3:44 PM
> > > > To: Salyzyn, Mark
> > > > Cc: Judith Lebzelter; James Bottomley; 
> > > > linux-scsi@vger.kernel.org; Duane Cox
> > > > Subject: Re: [PATCH] aacraid: Panics during init time reset 
> > > > (Was: [PATCH] aacraid: [Fastboot] Panics for AACRAID driver 
> > > > during 'insmod' for kexec test)
> > > > 
> > > > 
> > > > On Mon, Apr 02, 2007 at 02:34:36PM -0400, Salyzyn, Mark wrote:
> > > > > Duane discovered in the scsi-misc-2.6 code that the reset 
> > > > handler could
> > > > > be called without the sync command handler set up resulting 
> > > > in a panic.
> > > > > Judith discovered this issue within minutes and has 
> > > > recently reported
> > > > > it. Here is a fix.
> > > > 
> > > > Mark,
> > > > 
> > > > I applied this patch and ran a kexec test again and I still 
> > > > got a panic:
> > > > 
> > > > Loading aacraid.Adaptec aacraid driver (1.1-5[2437]-mh4)^M
> > > > ko module^M
> > > > ACPI: PCI Interrupt 0000:03:0e.0[A] -> Link [LNKC] -> GSI 3 
> > > > (level, low) -> IRQ 3^M
> > > > Unable to handle kernel NULL pointer dereference at 
> > > > 0000000000000000 RIP: ^M
> > > >  [<0000000000000000>]^M
> > > > PGD 4791067 PUD 473c067 PMD 0 ^M
> > > > Oops: 0010 [1] ^M
> > > > CPU 0 ^M
> > > > Modules linked in: aacraid^M
> > > > Pid: 977, comm: insmod Not tainted 2.6.21-rc3-kdump #1^M
> > > > RIP: 0010:[<0000000000000000>]  [<0000000000000000>]^M
> > > > RSP: 0000:ffff81000474dbf0  EFLAGS: 00010246^M
> > > > RAX: ffffc20000010000 RBX: ffff810004fe4cd8 RCX: 
> > 000000005b540e96^M
> > > > RDX: ffffc20000010000 RSI: ffff81000443cf40 RDI: 
> > ffff810004fe4cd8^M
> > > > RBP: 00000000fffee138 R08: ffffffff81001c20 R09: 
> > ffffffff8143593e^M
> > > > R10: ffff810004c537a0 R11: 0000000000000000 R12: 
> > ffff81000474dc7c^M
> > > > R13: 0000000000000000 R14: 0000000000000000 R15: 
> > 0000000000000000^M
> > > > FS:  000000000057b850(0063) GS:ffffffff814d6000(0000) 
> > > > knlGS:0000000000000000^M
> > > > CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b^M
> > > > CR2: 0000000000000000 CR3: 0000000004745000 CR4: 
> > 00000000000006e0^M
> > > > Process insmod (pid: 977, threadinfo ffff81000474c000, task 
> > > > ffff81000443cf40)^M
> > > > Stack:  ffffffff88008e82 00003e00fc1f0000 0000000000000000 
> > > > ffff810004fe4cd8^M
> > > >  ffff810004fe4800 0000000000000000 ffffffff8800a6dd 
> > 0000000000000032^M
> > > >  ffffffff88008c3b 0000000000000000 ffffffff00000000 
> > ffff81000474dc7c^M
> > > > Call Trace:^M
> > > >  [<ffffffff88008e82>] :aacraid:rx_sync_cmd+0x15c/0x16a^M
> > > >  [<ffffffff88008c3b>] :aacraid:aac_rx_restart_adapter+0x7e/0x169^M
> > > >  [<ffffffff88009121>] :aacraid:_aac_rx_init+0x7b/0x2fc^M
> > > >  [<ffffffff88002b1f>] :aacraid:aac_probe_one+0x1a2/0x457^M
> > > >  [<ffffffff8112be3d>] pci_device_probe+0x4c/0x75^M
> > > >  [<ffffffff8117d41e>] really_probe+0xc4/0x148^M
> > > >  [<ffffffff8117d64f>] __driver_attach+0x6d/0xab^M
> > > >  [<ffffffff8117d5e2>] __driver_attach+0x0/0xab^M
> > > >  [<ffffffff8117d5e2>] __driver_attach+0x0/0xab^M
> > > >  [<ffffffff8117c8f6>] bus_for_each_dev+0x43/0x6e^M
> > > >  [<ffffffff8117cc38>] bus_add_driver+0x6b/0x18d^M
> > > >  [<ffffffff8112bfdb>] __pci_register_driver+0x72/0xa7^M
> > > >  [<ffffffff8801203a>] :aacraid:aac_init+0x3a/0x75^M
> > > >  [<ffffffff8103bbcc>] sys_init_module+0x1195/0x12e6^M
> > > >  [<ffffffff8100913e>] system_call+0x7e/0x83^M
> > > > ^M
> > > > ^M
> > > > Code:  Bad RIP value.^M
> > > > RIP  [<0000000000000000>]^M
> > > >  RSP <ffff81000474dbf0>^M
> > > > CR2: 0000000000000000^M
> > > > 
> > > > There is an extra line in the call trace for the 'rx_sync_cmd'.
> > > > 
> > > > Judith
> > > > 
> > > > > 
> > > > > IMHO, this needs to be applied immediately regardless of 
> > > > the status of
> > > > > the kexec patch as this issue is present in the 
> > > > scsi-misc-2.6 driver for
> > > > > all existing init-time recovery actions. This patch in 
> > > > principal would
> > > > > not be different w/o the kexec patch.
> > > > > 
> > > > > ObligatoryDisclaimer: Please accept my condolences 
> > > > regarding Outlook's
> > > > > handling of patches.
> > > > > 
> > > > > This attached patch is against current scsi-misc-2.6
> > > > >  
> > > > > Signed-off-by: Mark Salyzyn <aacraid@adaptec.com>
> > > > > 
> > > > > ---
> > > > > 
> > > > > Sincerely -- Mark Salyzyn
> > > > 
> > > > 
> > > > 
> >