Re: [Qemu-devel] [PULL 00/10] SCSI patches for 2012-08-03

[Blue Swirl]

On Fri, Aug 3, 2012 at 8:37 PM, Anthony Liguori <address@hidden> wrote:
> Hervé Poussineau <address@hidden> writes:
>
>> Paolo Bonzini a écrit :
>>> Il 03/08/2012 21:27, Anthony Liguori ha scritto:
>>>> On 08/03/2012 03:06 AM, Paolo Bonzini wrote:
>>>>> Anthony,
>>>>>
>>>>> The following changes since commit
>>>>> 5e3bc7144edd6e4fa2824944e5eb16c28197dd5a:
>>>>>
>>>>>    Merge remote-tracking branch 'mst/tags/for_anthony' into staging
>>>>> (2012-07-30 10:00:48 -0500)
>>>>>
>>>>> are available in the git repository at:
>>>>>
>>>>>
>>>>>    git://github.com/bonzini/qemu.git scsi-next
>>>>>
>>>>> for you to fetch changes up to b71760ef6180662cc2dff1f6cf673a53508c59f1:
>>>>>
>>>>>    scsi: add support for ATA_PASSTHROUGH_xx scsi command (2012-08-03
>>>>> 10:04:37 +0200)
>>>>>
>>>>> The main change here is re-enabling megasas, but there's also Herve's
>>>>> retro HBA and a couple of random fixes.
>>>>   CC    sparc-softmmu/target-sparc/mmu_helper.o
>>>>   CC    sparc-softmmu/target-sparc/ldst_helper.o
>>>>   CC    sparc-softmmu/target-sparc/int32_helper.o
>>>>   LINK  sparc-softmmu/qemu-system-sparc
>>>> ../libhw64/hw/esp.o: In function `dc390_read_config':
>>>> /home/anthony/git/qemu/hw/esp.c:1220: undefined reference to
>>>> `pci_default_read_config'
>>>> ../libhw64/hw/esp.o: In function `dc390_write_config':
>>>> /home/anthony/git/qemu/hw/esp.c:1245: undefined reference to
>>>> `pci_default_write_config'
>>>> collect2: ld returned 1 exit status
>>>> make[1]: *** [qemu-system-sparc] Error 1
>>>> make: *** [subdir-sparc-softmmu] Error 2
>>>>
>>>> It looks like sparc does link in esp.c but doesn't support pci (at least
>>>> pci.o).  I must admit, it's not obvious to me whether pci.o should be
>>>> added to the sparc build or whether esp.c needs to be refactored.
>>>
>>> I think the latter.
>>>
>>> I removed this patch and pushed the rest (commit e4b6526) at the same
>>> location.  Blue/Herve, can you look at esp.c?
>>
>> I just sent an email to ML to add stubs for pci_default_read_config and
>> pci_default_write_config. I compile-tested sparc-softmmu target and it
>> works now well.
>
> Shouldn't Sparc consume pci.o?
>
> If Sparc doesn't have PCI, then stubs don't seem like the right answer.
> If it has PCI, then it ought to use the generic code, no?

Only JavaStation had PCI, but we don't emulate that.

Probably the same approach as lance & pcnet could be used.

>
> Regards,
>
> Anthony Liguori
>
>>
>> Hervé
>>
>> [1]
>