Re: [Qemu-devel] [PATCH 0/5] tests: add usb host adapter qtest cases

["Andreas Färber" <afaerber@suse.de>]

Hi Gonglei,

Am 19.06.2014 15:19, schrieb Gonglei (Arei):
>> -----Original Message-----
>> From: Andreas Färber [mailto:afaerber@suse.de]
>> Sent: Thursday, June 19, 2014 9:07 PM
>> To: Gerd Hoffmann
>> Cc: Gonglei (Arei); qemu-devel@nongnu.org; stefanha@redhat.com;
>> eblake@redhat.com; akong@redhat.com; Huangweidong (C); Luonengjun
>> Subject: Re: [PATCH 0/5] tests: add usb host adapter qtest cases
>>
>> Am 19.06.2014 14:47, schrieb Gerd Hoffmann:
>>> Maybe it makes sense to have a usb-hcd-hotplug-test.c file and collect
>>> hotplug testing for all usb hcds we have there.  For code sharing
>>> that'll be better for sure as the basic code flow is the same for all
>>> hcd types.
>>
>> Ah yes, seeing that there is a "global" QEMU instance in 4/5, that may
>> make more sense.
>>
> Hi, Gerd and Andreas,
> 
> Can we reach a consensus on this issue?

IMO it does not make sense to create a new ehci-hotplug file that will
only ever contain one test. Collecting all HCD hotplug tests into one
hcd-hotplug file, as Gerd suggested, would be fine with me. We'll still
need your hci-{ohci,uhci,xhci} files for functional (PCI/MMIO) tests and
what I referred to as "cold-plug" or init testing though.

>> OTOH hot-plug testing is a recurring topic for all PCI devices and won't
>> work in general (e.g., virtio-blk needs drive), 
> Yep.
> 
>> but maybe we can factor out some helpers to a .h file.
>>
> Such as... ?

Can't tell what Gerd had in mind for hcd hotplug sharing, but I was
thinking of some wrappers for the device_add and device_del commands.
Not sure if that makes sense in practice. Probably best start without
such a header and later we can see if there is much recurring code to be
factored out.

Regards,
Andreas

-- 
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg