Re: [PATCH v2 -rt] ide: workaround buggy hardware issues with preemptablehardirqs

[Sergei Shtylyov]

Hello.

Anton Vorontsov wrote:
> IDE interrupt handler relies on the fact that, if necessary, hardirqs
> will re-trigger on ISR exit. The assumption is valid for level sensitive
> interrupts.
>   

  It's valid for both edge and level triggered interrupts.

> But some hardware (namely ULi M5228 in the ULi M1575 "Super South Brige")
> behaves in a strange way: it asserts interrupts as edge sensitive. And
> because preemptable IRQ handler disables PIC's interrupt, PIC will likely
> miss it.
>   

  Unmasking an IRQ should re-enable an edge detector in a PIC (or that 
detector should even be independent from mask). So, unless an IRQ signal 
is a short pulse (which shouldn't be the case for IDE) it shouldn't be 
missed.

> This patch fixes following issue:
>
> ALI15X3: IDE controller (0x10b9:0x5229 rev 0xc8) at  PCI slot 0001:03:1f.0
> ALI15X3: 100% native mode on irq 18
> ide0: BM-DMA at 0x1120-0x1127, BIOS settings: hda:PIO, hdb:PIO
> ide1: BM-DMA at 0x1128-0x112f, BIOS settings: hdc:PIO, hdd:PIO
> hda: Optiarc DVD RW AD-7190A, ATAPI CD/DVD-ROM drive
> hda: UDMA/66 mode selected
> ide0 at 0x1100-0x1107,0x110a on irq 18
> ide-cd: cmd 0x5a timed out
> hda: lost interrupt
> hda: ATAPI 12X DVD-ROM DVD-R-RAM CD-R/RW drive, 2048kB Cache
> Uniform CD-ROM driver Revision: 3.20
> ide-cd: cmd 0x3 timed out
> hda: lost interrupt
> ide-cd: cmd 0x3 timed out
> hda: lost interrupt
> ...
>
> It would be great to re-configure the ULi bridge or ULi IDE controller
> to behave sanely, but no one knows how or if this is possible at all
> (no available specifications).
>
> So.. to workaround the issue IDE interrupt handler should re-check for
> any pending IRQs. This isn't bulletproof solution, but it works and this
> is the best one we can do

  I remembered a pssible issue with such code: it's possible that BSY 
bit will be cleared before INTRQ assertion (this is a race not spotted 
for a long time by the ATA standards -- and an observed behavior, at 
least of some ATAPI drives), thus a status register read won't clear the 
"interrupt pending" condition within drive.

MBR, Sergei


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