Skip to contentSkip to conversation

DMA problems with 3+ Promise IDE controllers

DISCUSSION Jaco van der Schyff 3 participants 3 messages Dec 23, 2004 – Dec 24, 2004

lore
Share & tools
lore
#0 lore raw

Hi,

I am curios if DMA bursting on the Promise cards are enabled by default on
2.6.x kernels.

CONFIG_PDC202XX_BURST exists for the old PDC driver, but there seems to be no similar option for the PDC202XX_NEW driver.

The problem I have is that I start to experience DMA errors as soon as I plug in a third Promise PDC20268 Ultra 100/TX2 card into one of my boxes. The system runs fine with only two cards, but as soon as the third card is connected and drives attached to it, the following errors are generated:

kernel: hde: max request size: 1024KiB
kernel: hde: 390721968 sectors (200049 MB) w/8192KiB Cache, CHS=24321/255/63, UD
MA(100)
kernel: hde:hde: dma_intr: status=0x51 { DriveReady SeekComplete Error }
kernel: hdg: max request size: 1024KiB
kernel: hdg: 390721968 sectors (200049 MB) w/8192KiB Cache, CHS=24321/255/63, UD
MA(100)
kernel: hdg:hdg: dma_intr: status=0x51 { DriveReady SeekComplete Error }

and

kernel: hde: dma_intr: error=0x84 { DriveStatusError BadCRC }
kernel: ide: failed opcode was: unknown
kernel: hde: dma_intr: status=0x51 { DriveReady SeekComplete Error }
kernel: hde: dma_intr: error=0x84 { DriveStatusError BadCRC }
kernel: ide: failed opcode was: unknown
kernel: hde: dma_intr: status=0x51 { DriveReady SeekComplete Error }
kernel: hde: dma_intr: error=0x84 { DriveStatusError BadCRC }
kernel: ide: failed opcode was: unknown
kernel: hde: dma_intr: status=0x51 { DriveReady SeekComplete Error }
kernel: hde: dma_intr: error=0x84 { DriveStatusError BadCRC }
kernel: ide: failed opcode was: unknown
kernel: PDC202XX: Primary channel reset.
kernel: ide2: reset: success
kernel:  hde1 hde2
kernel: hdg: dma_intr: error=0x84 { DriveStatusError BadCRC }
kernel: ide: failed opcode was: unknown
kernel: hdg: dma_intr: status=0x51 { DriveReady SeekComplete Error }
kernel: hdg: dma_intr: error=0x84 { DriveStatusError BadCRC }
kernel: ide: failed opcode was: unknown
kernel: hdg: dma_intr: status=0x51 { DriveReady SeekComplete Error }
kernel: hdg: dma_intr: error=0x84 { DriveStatusError BadCRC }
kernel: ide: failed opcode was: unknown
kernel: hdg: dma_intr: status=0x51 { DriveReady SeekComplete Error }
kernel: hdg: dma_intr: error=0x84 { DriveStatusError BadCRC }
kernel: ide: failed opcode was: unknown
kernel: PDC202XX: Secondary channel reset.
kernel: ide3: reset: success
kernel:  hdg1 hdg2

These errors are causing havok in my RAID5 setup.

IDE device detection in kernel:

kernel: ICH5: chipset revision 2
kernel: ICH5: not 100%% native mode: will probe irqs later
kernel: ide0: BM-DMA at 0xffa0-0xffa7, BIOS settings: hda:DMA,
hdb:pio
kernel: ide1: BM-DMA at 0xffa8-0xffaf, BIOS settings: hdc:DMA,
hdd:pio
kernel: PDC20268: IDE controller at PCI slot 0000:02:04.0
kernel: PCI: Found IRQ 9 for device 0000:02:04.0
kernel: PCI: Sharing IRQ 9 with 0000:00:1d.2
kernel: PCI: Sharing IRQ 9 with 0000:00:1f.1 helios kernel: PCI: Sharing IRQ 9 with 0000:01:01.0 helios kernel: PDC20268: chipset revision 2 helios kernel: PDC20268: ROM enabled at 0xff8c0000
kernel: PDC20268: 100%% native mode on irq 9
kernel: ide2: BM-DMA at 0x9400-0x9407, BIOS settings: hde:pio,
hdf:pio
kernel: ide3: BM-DMA at 0x9408-0x940f, BIOS settings: hdg:pio,
hdh:pio
kernel: PDC20268: IDE controller at PCI slot 0000:02:02.0
kernel: PCI: Found IRQ 12 for device 0000:02:02.0
kernel: PCI: Sharing IRQ 12 with 0000:00:1f.3
kernel: PDC20268: chipset revision 2
kernel: PDC20268: ROM enabled at 0xff8d0000
kernel: PDC20268: 100%% native mode on irq 12
kernel: ide4: BM-DMA at 0xa800-0xa807, BIOS settings: hdi:pio,
hdj:pio
kernel: ide5: BM-DMA at 0xa808-0xa80f, BIOS settings: hdk:pio,
hdl:pio
kernel: PDC20268: IDE controller at PCI slot 0000:02:00.0
kernel: PCI: Found IRQ 10 for device 0000:02:00.0
kernel: PDC20268: chipset revision 2
kernel: PDC20268: ROM enabled at 0xff8e0000
kernel: PDC20268: 100%% native mode on irq 10
kernel: ide6: BM-DMA at 0xbc00-0xbc07, BIOS settings: hdm:pio,
hdn:pio
kernel: ide7: BM-DMA at 0xbc08-0xbc0f, BIOS settings: hdo:pio,
hdp:pio

Any idea how I could get all three cards to work in UDMA100 mode?

Help will as always be greatly appreciated

- Jaco van der Schyff

#1 lore raw
Quoted 1 line · Jaco van der Schyff

Any idea how I could get all three cards to work in UDMA100 mode?

You either need multiple independant PCI busses or to write an arbitrator for the IDE DMA layer to ensure that no more than two of the three are doing DMA transfers at a time and the other waits.

Alan

#2 lore raw
Quoted 7 lines · Alan Cox
Quoted 1 line · Jaco van der Schyff

Any idea how I could get all three cards to work in UDMA100 mode?

You either need multiple independant PCI busses or to write an arbitrator for the IDE DMA layer to ensure that no more than two of the three are doing DMA transfers at a time and the other waits.

Could you explain why they can't coexist on the same PCI bus. The only problem I see with having all three cards DMA at the same time is increased latency. But he's getting CRC errors on the transfer between the card and the drive, and that doesn't seem like a latency issue.

I know Promise does funky stuff with their DMA, but I'm curious why no more than two cards can co-exist on the same PCI bus.

Ross