Re: [Qemu-devel] AHCI broken in current git, bisected.

[Alexey Zaytsev]

On Sun, May 15, 2011 at 21:14, Jan Kiszka <address@hidden> wrote:
> On 2011-05-15 18:32, Michael S. Tsirkin wrote:
>> On Sun, May 15, 2011 at 07:58:23PM +0400, Alexey Zaytsev wrote:
>>> Hi.
>>>
>>> The commit 667bb59d2358daeef179583c944becba3f1f9680
>>> Author: Avi Kivity <address@hidden>
>>> Date:   Mon Apr 4 18:28:02 2011 +0300
>>>
>>>     ich/ahci: convert to pci_register_bar_simple()
>>>
>>>     Signed-off-by: Avi Kivity <address@hidden>
>>>     Signed-off-by: Michael S. Tsirkin <address@hidden>
>>>
>>> breaks AHCI to the point the disks are not detected by both seabios and 
>>> Linux:
>>>
>>> [    8.582220] ata1: PATA max MWDMA2 cmd 0x1f0 ctl 0x3f6 bmdma 0xc000 irq 14
>>> [    8.582646] ata2: PATA max MWDMA2 cmd 0x170 ctl 0x376 bmdma 0xc008 irq 15
>>> [    8.603979] ACPI: PCI Interrupt Link [LNKD] enabled at IRQ 10
>>> [    8.605151] ahci 0000:00:04.0: PCI INT A -> Link[LNKD] -> GSI 10
>>> (level, high) -> IRQ 10
>>> [    9.608243] ahci 0000:00:04.0: controller reset failed (0xf000ff53)
>>> [    9.609948] ahci 0000:00:04.0: PCI INT A disabled
>>> [    9.610267] ahci: probe of 0000:00:04.0 failed with error -5
>>
>> Sure enough,
>>
>>     /* XXX BAR size should be 1k, but that breaks, so bump it to 4k for now 
>> */
>>     pci_register_bar_simple(&d->card, 5, 0x1000, 0, d->ahci.mem);
>>     msi_init(dev, 0x50, 1, true, false);
>>     ahci_init(&d->ahci, &dev->qdev, 6);
>>
>> where ahci_init initializes d->ahci.mem.
>>
>> Coul;d you try out the following please
>> (untested, a bit busy now)?
>
> See also http://permalink.gmane.org/gmane.comp.emulators.qemu/101978
>

Hey, thanks! I should look harder next time. ;)