qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [RESEND PULL 000/126] Buildsys / QOM / QDev / UI patches for 2023-02


From: Philippe Mathieu-Daudé
Subject: Re: [RESEND PULL 000/126] Buildsys / QOM / QDev / UI patches for 2023-02-27
Date: Mon, 27 Feb 2023 22:26:30 +0100
User-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:102.0) Gecko/20100101 Thunderbird/102.8.0

On 27/2/23 19:22, Peter Maydell wrote:
On Mon, 27 Feb 2023 at 14:08, Philippe Mathieu-Daudé <philmd@linaro.org> wrote:

----------------------------------------------------------------
- buildsys
   - Various header cleaned up (removing pointless headers)
   - Mark various files/code user/system specific
   - Make various objects target-independent
   - Remove tswapN() calls from dump.o
   - Suggest g_assert_not_reached() instead of assert(0)

- qdev / qom
   - Replace various container_of() by QOM cast macros
   - Declare some QOM macros using OBJECT_DECLARE_TYPE()
   - Embed OHCI QOM child in SM501 chipset

- hw (ISA & IDE)
   - add some documentation, improve function names
   - un-inline, open-code few functions
   - have ISA API accessing IRQ/DMA prefer ISABus over ISADevice
   - Demote IDE subsystem maintenance to "Odd Fixes"

- ui: Improve Ctrl+Alt hint on Darwin Cocoa

Seems to have broken the TCG plugins somehow?

https://gitlab.com/pm215/qemu/-/jobs/3841174348

TEST munmap-pthread on aarch64
**
ERROR:../plugins/core.c:221:qemu_plugin_vcpu_init_hook: assertion
failed: (success)

Similarly in
https://gitlab.com/pm215/qemu/-/jobs/3841174344
https://gitlab.com/pm215/qemu/-/jobs/3841174341

Sorry since this wasn't reliable, I first thought this was
pre-existing / unrelated to this patchset. But...

012add0d52621e34632ea24361a0c751bc2f5ad1 is the first bad commit
commit 012add0d52621e34632ea24361a0c751bc2f5ad1
Author: Philippe Mathieu-Daudé <philmd@linaro.org>
Date:   Tue Dec 6 16:06:00 2022 +0100

    cputlb: Restrict SavedIOTLB to system emulation

    Commit 2f3a57ee47 ("cputlb: ensure we save the IOTLB data in
    case of reset") added the SavedIOTLB structure -- which is
    system emulation specific -- in the generic CPUState structure.

For some reason this was hard to bisect. The usual "make check-tcg"
rule wasn't coherent along bisection. Eventually I ended using
"ninja clean && ninja qemu-user && make check-tcg" which worked.

Could a .so rule be incomplete, so while qemu-user binary is rebuilt,
the plugin sharedlibs aren't?

Anyhow, short term I'll simply drop this commit from the pullreq.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]