|
From: | Philippe Mathieu-Daudé |
Subject: | Re: [PATCH v2 0/8] hw: Convert various reset() handler to DeviceReset |
Date: | Thu, 10 Oct 2019 15:08:03 +0200 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.1.0 |
On 10/10/19 3:05 AM, Li Qiang wrote:
Philippe Mathieu-Daudé <address@hidden <mailto:address@hidden>> 于 2019年10月10日周四 上午3:54写道:Hi Li, On 10/9/19 4:28 AM, Li Qiang wrote: > Philippe Mathieu-Daudé <address@hidden <mailto:address@hidden> <mailto:address@hidden <mailto:address@hidden>>> 于 > 2019年10月8日周二 下午10:47写道: > > Since v1: > - Removed the pci-host devices > > > Hello I want to know why remove this? I haven't removed the devices, I simply remove the patches convertingthem to DeviceReset,Yes, I mean the patches. basically because I've not enough time to check ifthe are on a bus that would reset them.IIUC, they are right. I added these devices on my TODO list for later, so meanwhile the other devices can be easily reviewed and merged. When few patches from a series are not reviewed or incorrect, sometime the rest of the series is not merged, so I prefer to split it and get these patches merged.As far as I can see, most of the devices' usage of qemu_register_reset function can beconvert to 'dc->reset'. In the main function. qemu_register_reset(qbus_reset_all_fn, sysbus_get_default());The 'qbus_reset_all_fn' calls 'qbus_reset_all' from the 'main-sys-bus'. Then 'qdev_reset_one' will call 'device_reset'. So IIUC every bus attached to 'main-sys-bus' can be reset through 'dc->reset'So I'm quite sure most of the cases that devices use 'qemu_register_reset' can be changed to 'dc->reset'. Seems you're busy, If you don't mind, I can do some of the work to convert 'reset' callback(not a patchset, one by one).
I guess we are all busy ;) I'm just trying to prioritize here.This series is not very important because what we have today works, and I would rather not introduce regressions. What happened then is it is the 3rd time at least I get confuse with the qemu_register_reset() function while reviewing code. Then my rule of thumb is if an improvement takes less than 1h, then I just do it and keep going, else if I postpone it I'll never go back to it. When a series start to take too much rework it means I might started it wrongly, or I underestimated the time required. Time that is taken of other more important tasks. Today I prefer merged a subset of the series that is correct, rather than aiming for the whole and never get it merged.
If you are interested in respining/fixing the pci-host devices I'd be very thankful! I appreciate your help (and reviews).
Regards, Phil. PD: I'll respin as v3 with your tags and the PIIX4_IDE fix.
> > - Removed the vmcoreinfo conversion (elmarco) but add a comment. > - Added Igor's R-b tag. > > Following the thread discussion between Peter/Markus/Damien about > reset handlers: > https://www.mail-archive.com/address@hidden/msg617103.html > I started to remove qemu_register_reset() calls from few qdevified > devices (the trivial ones). > > Regards, > > Phil. > > v1: https://lists.gnu.org/archive/html/qemu-devel/2019-09/msg06367.html > > Philippe Mathieu-Daudé (8): > hw/acpi/piix4: Convert reset handler to DeviceReset > hw/isa/piix4: Convert reset handler to DeviceReset > hw/ide/piix: Convert reset handler to DeviceReset > hw/ide/sii3112: Convert reset handler to DeviceReset > hw/ide/via82c: Convert reset handler to DeviceReset > hw/isa/vt82c686: Convert reset handler to DeviceReset > hw/input/lm832x: Convert reset handler to DeviceReset > hw/misc/vmcoreinfo: Document its reset handler
[Prev in Thread] | Current Thread | [Next in Thread] |