On 11/11/2011 05:11 PM, Matthew Garrett wrote:
> On Fri, Nov 11, 2011 at 12:27:55PM +0000, James Bottomley wrote:
> 
>> The next question is: is the driver the correct place?  This sounds like
>> a PCIe Link Power Management blacklist set ... which might need updating
>> on the fly ... might we need a user knob for this (like we have for the
>> SCSI black/white list)?
> 
> The only thing that knows whether any given piece of hardware is broken 
> here is the hardware-specific driver. There's already a user knob for 
> this, but some devices will fall over the moment the hardware is enabled 
> if the ASPM state is broken which makes it more difficult to fix up that 
> way.

It's quite hard to identify all affected hardware, as far as I understand in all cases it was various controllers.
Probably hardware vendor could do it, but I doubt that it is possible in current situation.

Just some details about my test node:
it was Adaptec 5805

[    3.045666] Adaptec aacraid driver 1.1-7[28000]-ms
[    3.045964] aacraid 0000:04:00.0: PCI INT A -> GSI 32 (level, low) -> IRQ 32
[    3.311288] AAC0: kernel 5.2-0[18252] Nov 22 2010
[    3.311398] AAC0: monitor 5.2-0[18252]
[    3.311508] AAC0: bios 5.2-0[18252]
[    3.311611] AAC0: serial 1D2811B07D0
[    3.311715] AAC0: Non-DASD support enabled.
[    3.311818] AAC0: 64bit support enabled.
[    3.311927] AAC0: 64 Bit DAC enabled
[    3.319115] scsi0 : aacraid

When  ASPM policy was changed on non-patched driver there was lot of following messages in dmesg

[    6.279007] pcieport 0000:00:09.0:    [ 6] Bad TLP               
[    6.279119] pcieport 0000:00:09.0: AER: Corrected error received: id=0048
[    6.279378] pcieport 0000:00:09.0: AER: Corrected error received: id=0048
[    6.279495] pcieport 0000:00:09.0: PCIe Bus Error: severity=Corrected, type=Data Link Layer, id=0048(Receiver ID)
[    6.279661] pcieport 0000:00:09.0:   device [8086:3410] error status/mask=00000040/00000001
[    6.279823] pcieport 0000:00:09.0:    [ 6] Bad TLP               
[    6.283044] pcieport 0000:00:09.0: AER: Corrected error received: id=0048
[    6.283185] pcieport 0000:00:09.0: PCIe Bus Error: severity=Corrected, type=Data Link Layer, id=0048(Receiver ID)
[    6.283391] pcieport 0000:00:09.0:   device [8086:3410] error status/mask=00000040/00000001

lspci output is attached

thank you,
	Vasily Averin