Re: [PATCH v3 00/20] hw/i386/pc: Split PIIX3 southbridge from i440FX northbridge

["Philippe Mathieu-Daudé" <philmd@redhat.com>]

Hi Aleksandar,

On 10/27/19 8:44 AM, Aleksandar Markovic wrote:
> On Saturday, October 26, 2019, Philippe Mathieu-Daudé <philmd@redhat.com 
> <mailto:philmd@redhat.com>> wrote:
> 
>     Changes since v2 [0]:
>     - Use a #define
>     - Reword one description
>     - Added review tags (thanks all for reviewing!)
> 
>     Changes since v1 [1]:
>     - Removed patch reintroducing DO_UPCAST() use (thuth)
>     - Took various patches out to reduce series (thuth)
>     - Added review tags (thanks all for reviewing!)
> 
>     $ git backport-diff -u pc_split_i440fx_piix-v2
>     Key:
>     [----] : patches are identical
>     [####] : number of functional differences between
>     upstream/downstream patch
>     [down] : patch is downstream-only
>     The flags [FC] indicate (F)unctional and (C)ontextual differences,
>     respectively
> 
>     001/20:[----] [--] 'MAINTAINERS: Keep PIIX4 South Bridge separate
>     from PC Chipsets'
>     002/20:[0004] [FC] 'piix4: Add the Reset Control Register'
>     003/20:[0002] [FC] 'piix4: Add an i8259 Interrupt Controller as
>     specified in datasheet'
>     004/20:[----] [--] 'Revert "irq: introduce qemu_irq_proxy()"'
>     005/20:[----] [--] 'piix4: Rename PIIX4 object to piix4-isa'
>     006/20:[----] [--] 'piix4: Add an i8257 DMA Controller as specified
>     in datasheet'
>     007/20:[----] [-C] 'piix4: Add an i8254 PIT Controller as specified
>     in datasheet'
>     008/20:[0004] [FC] 'piix4: Add a MC146818 RTC Controller as
>     specified in datasheet'
>     009/20:[----] [--] 'hw/mips/mips_malta: Create IDE hard drive array
>     dynamically'
>     010/20:[----] [--] 'hw/mips/mips_malta: Extract the PIIX4 creation
>     code as piix4_create()'
>     011/20:[----] [-C] 'hw/isa/piix4: Move piix4_create() to hw/isa/piix4.c'
>     012/20:[----] [--] 'hw/i386: Remove obsolete
>     LoadStateHandler::load_state_old handlers'
>     013/20:[----] [--] 'hw/pci-host/piix: Extract piix3_create()'
>     014/20:[0002] [FC] 'hw/pci-host/piix: Move RCR_IOPORT register
>     definition'
>     015/20:[----] [--] 'hw/pci-host/piix: Define and use the PIIX IRQ
>     Route Control Registers'
>     016/20:[----] [-C] 'hw/pci-host/piix: Move i440FX declarations to
>     hw/pci-host/i440fx.h'
>     017/20:[----] [--] 'hw/pci-host/piix: Fix code style issues'
>     018/20:[----] [--] 'hw/pci-host/piix: Extract PIIX3 functions to
>     hw/isa/piix3.c'
>     019/20:[----] [--] 'hw/pci-host: Rename incorrectly named 'piix' as
>     'i440fx''
>     020/20:[0004] [FC] 'hw/pci-host/i440fx: Remove the last PIIX3 traces'
> 
>     Previous cover:
> 
>     This series is a rework of "piix4: cleanup and improvements" [2]
>     from Hervé, and my "remove i386/pc dependency: PIIX cleanup" [3].
> 
>     Still trying to remove the strong X86/PC dependency 2 years later,
>     one step at a time.
>     Here we split the PIIX3 southbridge from i440FX northbridge.
>     The i440FX northbridge is only used by the PC machine, while the
>     PIIX southbridge is also used by the Malta MIPS machine.
> 
>     This is also a step forward using KConfig with the Malta board.
>     Without this split, it was impossible to compile the Malta without
>     pulling various X86 pieces of code.
> 
>     The overall design cleanup is not yet perfect, but enough to post
>     as a series.
> 
>     Now that the PIIX3 code is extracted, the code duplication with the
>     PIIX4 chipset is obvious. Not worth improving for now because it
>     isn't broken.
> 
>     Based-on: <1572097538-18898-1-git-send-email-pbonzini@redhat.com
>     <mailto:1572097538-18898-1-git-send-email-pbonzini@redhat.com>>
>     to include:
>     mc146818rtc: Allow call object_initialize(MC146818_RTC) instead of
>     rtc_init()
>     https://mid.mail-archive.com/20191018133547.10936-1-philmd@redhat.com <https://mid.mail-archive.com/20191018133547.10936-1-philmd@redhat.com>
> 
>     Since Aleksandar offered me to send the pull request [4] I plan to do
>     it once Paolo's pull [5] is merged.
> 
> 
> Philippe,
> 
> I attempted the other day the integration of v2 of this series into MIPS 
> pull request, but couldn't do it - since another series of yours was 
> already merged, acting on the same code, making rebasing difficult. Now 
> this, v3, series can't be applied since certain patches in some, on 
> surface, unrelated series aren't megred, and v3 assumes they are merged.
> 
> If you send a series, it should preferably be based on the latest 
> (current) code base, not on some imagined future state.

I used the 'based-on' tag to refer other series, and patchew succeeded
at applying this series on top of it and build/test it.

Based-on: <1572097538-18898-1-git-send-email-pbonzini@redhat.com

> Why did you create this such mess with interdependencies of your own 
> multiple series, and just right before softfreeze? :( You should have 
> distributed submitting those series over longer time interval, and 
> absolutely avoid, if possible, this hectic around-softfreeze period. You 
> did the opposite: waited for softfreeze to become close, and sent 
> several interdependant series in matter of days - creating stress 
> without any real technical reason.

This series touches multiple subsystems, so different maintainers had to 
take the various parts, so I had to split.

Also Peter gave the recommendation on the list to not put more than 20 
patches in a series, to make it digestible for review.

> In case you, for any reason, can't complete this by softfreeze, I advice 
> you not to rush, and postpone the integration to 5.0.

This series doesn't provide any useful feature, it is a simple cleanup,
posted and reviewed before soft freeze, so we still have 1 week (until
hard freeze) to have it merged, or postpone. No need to stress out for
a cleanup ;)

Regards,

Phil.