Message ID | 20190507163416.24647-1-philmd@redhat.com (mailing list archive) |
---|---|
Headers | show |
Series | hw: Use object_initialize_child for correct reference counting | expand |
Hi Eduardo, On 5/7/19 6:34 PM, Philippe Mathieu-Daudé wrote: > Hi, > > This series looks at Eduardo suggestions from [1] > and Thomas commit aff39be0ed97 to replace various > object_initialize + qdev_set_parent_bus calls by > sysbus_init_child_obj(). Do you think you can take this series? Else, via which tree it should go? Thanks! Phil. > > Important comment from Eduardo: > > It's possible, but we need a volunteer to review each > hunk because the existing code might be (correctly) > calling object_unref() (either immediately or when > parent is finalized). > > I tried to split it enough to make the review process > easier. > > Regards, > > Phil. > > [*] https://patchwork.ozlabs.org/patch/943333/#1953608 > v1: https://lists.gnu.org/archive/html/qemu-devel/2019-02/msg05931.html > > Philippe Mathieu-Daudé (16): > hw/ppc/pnv: Use object_initialize_child for correct reference counting > hw/misc/macio: Use object_initialize_child for correct ref. counting > hw/virtio: Use object_initialize_child for correct reference counting > hw/arm/bcm2835: Use TYPE_PL011 instead of hardcoded string > hw/arm/bcm2835: Use object_initialize() on PL011State > hw/arm/bcm2835: Use object_initialize_child for correct ref. counting > hw/arm/aspeed: Use object_initialize_child for correct ref. counting > hw/arm: Use object_initialize_child for correct reference counting > hw/mips: Use object_initialize() on MIPSCPSState > hw/mips: Use object_initialize_child for correct reference counting > hw/microblaze/zynqmp: Move the IPI state into the PMUSoC state > hw/microblaze/zynqmp: Let the SoC manage the IPI devices > hw/microblaze/zynqmp: Use object_initialize_child for correct ref. > counting > hw/microblaze/zynqmp: Use object_initialize_child for correct ref. > counting > hw/arm/mps2: Use object_initialize_child for correct reference > counting > hw/intc/nvic: Use object_initialize_child for correct reference > counting
On Fri, May 17, 2019 at 12:32:18PM +0200, Philippe Mathieu-Daudé wrote: > Hi Eduardo, > > On 5/7/19 6:34 PM, Philippe Mathieu-Daudé wrote: > > Hi, > > > > This series looks at Eduardo suggestions from [1] > > and Thomas commit aff39be0ed97 to replace various > > object_initialize + qdev_set_parent_bus calls by > > sysbus_init_child_obj(). > > Do you think you can take this series? > Else, via which tree it should go? I was expecting the maintainers of each architecture to apply the patches for their areas. But I'd be glad to merge it through my tree if it makes it easier for everybody. Are the arm, microblaze, mips, and ppc maintainers OK with that? > > Thanks! > > Phil. > > > > > Important comment from Eduardo: > > > > It's possible, but we need a volunteer to review each > > hunk because the existing code might be (correctly) > > calling object_unref() (either immediately or when > > parent is finalized). > > > > I tried to split it enough to make the review process > > easier. > > > > Regards, > > > > Phil. > > > > [*] https://patchwork.ozlabs.org/patch/943333/#1953608 > > v1: https://lists.gnu.org/archive/html/qemu-devel/2019-02/msg05931.html > > > > Philippe Mathieu-Daudé (16): > > hw/ppc/pnv: Use object_initialize_child for correct reference counting > > hw/misc/macio: Use object_initialize_child for correct ref. counting > > hw/virtio: Use object_initialize_child for correct reference counting > > hw/arm/bcm2835: Use TYPE_PL011 instead of hardcoded string > > hw/arm/bcm2835: Use object_initialize() on PL011State > > hw/arm/bcm2835: Use object_initialize_child for correct ref. counting > > hw/arm/aspeed: Use object_initialize_child for correct ref. counting > > hw/arm: Use object_initialize_child for correct reference counting > > hw/mips: Use object_initialize() on MIPSCPSState > > hw/mips: Use object_initialize_child for correct reference counting > > hw/microblaze/zynqmp: Move the IPI state into the PMUSoC state > > hw/microblaze/zynqmp: Let the SoC manage the IPI devices > > hw/microblaze/zynqmp: Use object_initialize_child for correct ref. > > counting > > hw/microblaze/zynqmp: Use object_initialize_child for correct ref. > > counting > > hw/arm/mps2: Use object_initialize_child for correct reference > > counting > > hw/intc/nvic: Use object_initialize_child for correct reference > > counting
On Fri, 17 May 2019 at 18:56, Eduardo Habkost <ehabkost@redhat.com> wrote: > > On Fri, May 17, 2019 at 12:32:18PM +0200, Philippe Mathieu-Daudé wrote: > > Hi Eduardo, > > > > On 5/7/19 6:34 PM, Philippe Mathieu-Daudé wrote: > > > Hi, > > > > > > This series looks at Eduardo suggestions from [1] > > > and Thomas commit aff39be0ed97 to replace various > > > object_initialize + qdev_set_parent_bus calls by > > > sysbus_init_child_obj(). > > > > Do you think you can take this series? > > Else, via which tree it should go? > > I was expecting the maintainers of each architecture to apply the > patches for their areas. This in my experience rarely happens, because splitting up a patchset is effort and there's a coordination problem working out who's going to take which patches -- it's why it works better to have several series each of which covers one submaintainer's area, rather than one big series which then doesn't have an obvious path into the tree. (Personally I also tend to treat omnibus patch series as "somebody else's problem" whereas patch series that are mostly or entirely arm changes go on my list as needing to be dealt with...) > Are the arm, microblaze, mips, and ppc maintainers OK with that? No objections for arm if the patches have been reviewed. thanks -- PMM
> > On 5/7/19 6:34 PM, Philippe Mathieu-Daudé wrote: > > > Hi, > > > > > > This series looks at Eduardo suggestions from [1] > > > and Thomas commit aff39be0ed97 to replace various > > > object_initialize + qdev_set_parent_bus calls by > > > sysbus_init_child_obj(). > > > > Do you think you can take this series? > > Else, via which tree it should go? > > I was expecting the maintainers of each architecture to apply the > patches for their areas. But I'd be glad to merge it through my > tree if it makes it easier for everybody. > > Are the arm, microblaze, mips, and ppc maintainers OK with that? Hello, Eduardo. I am OK with two patches applicable to MIPS. Moreover, I am going to apply them to my pull request scheduled to be sent today. Sorry if that makes your part more difficult (you will have to skip two patches from this series). The reason for my urgency is that we in Wave start regression testing for QEMU 4.1 in MIPS environments, and we wanted these two patches integrated sooner rather than later. Thanks to everyone involved! Aleksandar