[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-block] [Qemu-devel] [PATCH RFC for-2.3? 0/8] prep: Fix pc87312
Re: [Qemu-block] [Qemu-devel] [PATCH RFC for-2.3? 0/8] prep: Fix pc87312 for -device usage
Mon, 30 Mar 2015 18:42:20 +0200
Gnus/5.13 (Gnus v5.13) Emacs/24.3 (gnu/linux)
Paolo Bonzini <address@hidden> writes:
> On 30/03/2015 16:25, Markus Armbruster wrote:
>> Andreas Färber <address@hidden> writes:
>>> Hello Markus et al.,
>>> This series attempts to fix the -device pc87312 issues you reported.
>>> I can't add alias properties for devices that don't get created
>>> before realize.
>>> Therefore this involves moving code for various ISA devices, to enable us
>>> to initialize the objects early for alias properties and realizing them
>>> as part of the composite device once the configuration is known, also fixing
>>> error propagation while at it. Probably needs a further iteration.
>>> Yes, it's terribly invasive, that's why the code was as it is.
>>> But the code movements are quite trivial, as long as no in-air
>>> conflicts occur.
>> Pity we have to move the state structs to the header, but that what we
>> have to do to make them embeddable. And I understand embedding is how
>> we do sub-devices ("part of" instead of "plugged into").
>> Your changes are quite regular. They don't look scary to me, but that
>> could be just ignorance. They're non-trivial enough though to make wary
>> of merging them for 2.3 this late.
> The question is really: what is gained from this series as of 2.3? Is
> anything actually using "-device pc87312"? It was broken before IIUC,
> so it can remain broken for one more version. Markus's patches would
> just add one more "git revert" to this series, basically.
My series adds four FIXME comments in two commits. Both commits add
more of them elsewhere.
If my series goes in first, then rebasing this series on top should drop
the four FIXMEs again.
If this series goes first, then I have to rebase minr to drop the four
FIXMEs and replace an example in a commit message.