[2.6.30.1] DMA debug/AHCI SG inconsistency...

[Daniel J Blueman]

When running 2.6.30.1 checked for a while, I observed:

WARNING: at lib/dma-debug.c:565 check_unmap+0x518/0x5f0()

Hardware name: OEM

ahci 0000:00:1f.2: DMA-API: device driver frees DMA sg list with
different entry count [map count=124] [unmap count=117]

Modules linked in: coretemp w83627ehf hwmon_vid snd_hda_codec_realtek
snd_hda_intel snd_hda_codec snd_pcm snd_timer ath9k snd mac80211
led_class soundcore snd_page_alloc pl2303

Pid: 0, comm: swapper Not tainted 2.6.30.1-273s #1

Call Trace:

 <IRQ>  [<ffffffff803f6b98>] ? check_unmap+0x518/0x5f0

 [<ffffffff8023ffb8>] warn_slowpath_common+0x78/0xd0

 [<ffffffff80240094>] warn_slowpath_fmt+0x64/0x70

 [<ffffffff805ca430>] ? _spin_lock_irqsave+0x70/0x90

 [<ffffffff803f678c>] ? check_unmap+0x10c/0x5f0

 [<ffffffff803f6b98>] check_unmap+0x518/0x5f0

 [<ffffffff80266675>] ? lock_release_holdtime+0x35/0x1c0

 [<ffffffff803f6e2d>] debug_dma_unmap_sg+0x12d/0x1e0

 [<ffffffff804794c7>] ata_sg_clean+0x77/0xf0

 [<ffffffff80479575>] __ata_qc_complete+0x35/0x110

 [<ffffffff8047970d>] ata_qc_complete+0xbd/0x250

 [<ffffffff80479c47>] ata_qc_complete_multiple+0x97/0xe0

 [<ffffffff8048ee74>] ahci_interrupt+0x94/0x550

 [<ffffffff80266675>] ? lock_release_holdtime+0x35/0x1c0

 [<ffffffff8027a5da>] handle_IRQ_event+0x3a/0xe0

 [<ffffffff8027c291>] handle_edge_irq+0xc1/0x160

 [<ffffffff8020efc6>] handle_irq+0x46/0xa0

 [<ffffffff8020e42a>] do_IRQ+0x6a/0xf0

 [<ffffffff8020c793>] ret_from_intr+0x0/0xf

 <EOI>  [<ffffffff80213e25>] ? mwait_idle+0x75/0x90

 [<ffffffff80213e1c>] ? mwait_idle+0x6c/0x90

 [<ffffffff8020b030>] ? cpu_idle+0xa0/0x140

 [<ffffffff805bc6d5>] ? rest_init+0x65/0x70

 [<ffffffff807bbb9d>] ? start_kernel+0x39a/0x449

 [<ffffffff807bb0a9>] ? x86_64_start_reservations+0x99/0xb9

 [<ffffffff807bb1a9>] ? x86_64_start_kernel+0xe0/0xf2

---[ end trace df6d791d18fe80b5 ]---

Mapped at:

 [<ffffffffffffffff>] 0xffffffffffffffff

(then ~4000s later)
DMA-API: debugging out of memory - disabling

The logic in the ata_sg_setup->dma_map_sg->debug_dma_map_sg and
ata_sg_clean->dma_unmap_sg->debug_dma_unmap_sg->check_unmap call
chains looks good, perhaps pointing to data corruption/race in the DMA
debug API or problem with the AHCI driver?

Daniel
-- 
Daniel J Blueman
--
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/