[Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Thomas Huth]

These files lack an entry in the MAINTAINERS file, and according to
the initial commits, the board and devices are incomplete. Since there
have hardly been any commits in the past to really improve them, we
should consider to mark them as deprecated now.

Thomas Huth (2):
  hw/arm: Deprecate the "collie" board
  arm: Deprecate the Strongarm sa1100 and sa1110 processors

 hw/arm/collie.c      |  1 +
 qemu-deprecated.texi | 10 ++++++++++
 2 files changed, 11 insertions(+)

-- 
1.8.3.1

[Qemu-devel] [PATCH 1/2] hw/arm: Deprecate the "collie" board

[Thomas Huth]

"collie" has no entry in the MAINTAINERS file, and the initial commit
with ID c64b21d519a6ecae12f65625fa60f3035ed88644 said:

"Add very basic implementation of collie PDA emulation. The system lacks
 LoCoMo and graphics/sound emulation. Linux kernel boots up to mounting
 rootfs (theoretically it can be provided in pflash images)."

Thus the board is incomplete, since after that initial commit, there were
only the usual QEMU API-related rework patches applied to this file.
Unless someone speaks up and says that this board is still useful, we
should consider to remove it, so mark it as deprecated now.

Signed-off-by: Thomas Huth <thuth@redhat.com>
---
 hw/arm/collie.c      | 1 +
 qemu-deprecated.texi | 5 +++++
 2 files changed, 6 insertions(+)

diff --git a/hw/arm/collie.c b/hw/arm/collie.c
index 48b732c..e41bbc5 100644
--- a/hw/arm/collie.c
+++ b/hw/arm/collie.c
@@ -60,6 +60,7 @@ static void collie_machine_init(MachineClass *mc)
     mc->init = collie_init;
     mc->ignore_memory_transaction_failures = true;
     mc->default_cpu_type = ARM_CPU_TYPE_NAME("sa1110");
+    mc->deprecation_reason = "board is incomplete and unmaintained";
 }
 
 DEFINE_MACHINE("collie", collie_machine_init)
diff --git a/qemu-deprecated.texi b/qemu-deprecated.texi
index 11b870c..acf9809 100644
--- a/qemu-deprecated.texi
+++ b/qemu-deprecated.texi
@@ -129,6 +129,11 @@ This machine type uses an unmaintained firmware, broken in lots of ways,
 and unable to start post-2004 operating systems. 40p machine type should be
 used instead.
 
+@subsection collie (ARM) (since 3.1)
+
+The board is incomplete and unmaintained. Use a different ARM board if
+possible instead.
+
 @section Device options
 
 @subsection Block device options
-- 
1.8.3.1

[Qemu-devel] [PATCH 2/2] arm: Deprecate the Strongarm sa1100 and sa1110 processors

[Thomas Huth]

The deprecated "collie" board is the only user of the Strongarm
devices, so if "collie" goes away, we should remove the Strongarm
devices, too.

Signed-off-by: Thomas Huth <thuth@redhat.com>
---
 qemu-deprecated.texi | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/qemu-deprecated.texi b/qemu-deprecated.texi
index acf9809..0de5c7f 100644
--- a/qemu-deprecated.texi
+++ b/qemu-deprecated.texi
@@ -116,6 +116,11 @@ The @option{[hub_id name]} parameter tuple of the 'hostfwd_add' and
 The ``ivshmem'' device type is replaced by either the ``ivshmem-plain''
 or ``ivshmem-doorbell`` device types.
 
+@subsection Strongarm sa1100 and sa1110 and related devices (since 3.1)
+
+Without the deprecated "collie" board there is no other machine which is
+able to use these devices.
+
 @section System emulator machines
 
 @subsection pc-0.10 and pc-0.11 (since 3.0)
-- 
1.8.3.1

Re: [Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Peter Maydell]

On 26 October 2018 at 11:06, Thomas Huth <thuth@redhat.com> wrote:
> These files lack an entry in the MAINTAINERS file, and according to
> the initial commits, the board and devices are incomplete. Since there
> have hardly been any commits in the past to really improve them, we
> should consider to mark them as deprecated now.
>
> Thomas Huth (2):
>   hw/arm: Deprecate the "collie" board
>   arm: Deprecate the Strongarm sa1100 and sa1110 processors

Hi Guenter; there's a proposal here to deprecate (and eventually
remove) the 'collie' board (strongarm) from QEMU. Is that one of
the ones you're currently using in your automated testing of Linux
kernels on QEMU?

thanks
-- PMM

Re: [Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Guenter Roeck]

On 10/26/18 3:12 AM, Peter Maydell wrote:
> On 26 October 2018 at 11:06, Thomas Huth <thuth@redhat.com> wrote:
>> These files lack an entry in the MAINTAINERS file, and according to
>> the initial commits, the board and devices are incomplete. Since there
>> have hardly been any commits in the past to really improve them, we
>> should consider to mark them as deprecated now.
>>
>> Thomas Huth (2):
>>    hw/arm: Deprecate the "collie" board
>>    arm: Deprecate the Strongarm sa1100 and sa1110 processors
> 
> Hi Guenter; there's a proposal here to deprecate (and eventually
> remove) the 'collie' board (strongarm) from QEMU. Is that one of
> the ones you're currently using in your automated testing of Linux
> kernels on QEMU?
> 

Yes. I can run the test with older versions of qemu, so it is ok for me
if it is removed (as long as that removal is not backported).

Guenter

Re: [Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Peter Maydell]

On 27 October 2018 at 12:04, Guenter Roeck <linux@roeck-us.net> wrote:
> On 10/26/18 3:12 AM, Peter Maydell wrote:
>> Hi Guenter; there's a proposal here to deprecate (and eventually
>> remove) the 'collie' board (strongarm) from QEMU. Is that one of
>> the ones you're currently using in your automated testing of Linux
>> kernels on QEMU?
>>
>
> Yes. I can run the test with older versions of qemu, so it is ok for me
> if it is removed (as long as that removal is not backported).

Mmm, but if we have an active user who's testing them then they
probably shouldn't be in the frontline of boards to remove.
Which other boards do you test with mainline QEMU?

thanks
-- PMM

Re: [Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Guenter Roeck]

On 10/29/18 6:24 AM, Peter Maydell wrote:
> On 27 October 2018 at 12:04, Guenter Roeck <linux@roeck-us.net> wrote:
>> On 10/26/18 3:12 AM, Peter Maydell wrote:
>>> Hi Guenter; there's a proposal here to deprecate (and eventually
>>> remove) the 'collie' board (strongarm) from QEMU. Is that one of
>>> the ones you're currently using in your automated testing of Linux
>>> kernels on QEMU?
>>>
>>
>> Yes. I can run the test with older versions of qemu, so it is ok for me
>> if it is removed (as long as that removal is not backported).
> 
> Mmm, but if we have an active user who's testing them then they
> probably shouldn't be in the frontline of boards to remove.
> Which other boards do you test with mainline QEMU?
> 

For arm:

akita
ast2500-evb
beagle
beaglexm
borzoi
collie
cubieboard
imx25-pdk
integratorcp
kzm
mainstone
midway
mps2-an385
overo
palmetto-bmc
raspi2
realview-eb
realview-eb-mpcore
realview-pb-a8
realview-pbx-a9
romulus-bmc
sabrelite
smdkc210
spitz
terrier
tosa
versatileab
versatilepb
vexpress-a15
vexpress-a15-a7
vexpress-a9
witherspoon-bmc
xilinx-zynq-a9
z2

Though not all of them are supported by upstream qemu. For some of them I carry local patches,
for others I use out-of-tree versions of qemu (beagle/beaglexm).

Guenter

Re: [Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Philippe Mathieu-Daudé]

Hi Guenter,

On 29/10/18 15:09, Guenter Roeck wrote:
> On 10/29/18 6:24 AM, Peter Maydell wrote:
>> On 27 October 2018 at 12:04, Guenter Roeck <linux@roeck-us.net> wrote:
>>> On 10/26/18 3:12 AM, Peter Maydell wrote:
>>>> Hi Guenter; there's a proposal here to deprecate (and eventually
>>>> remove) the 'collie' board (strongarm) from QEMU. Is that one of
>>>> the ones you're currently using in your automated testing of Linux
>>>> kernels on QEMU?
>>>>
>>>
>>> Yes. I can run the test with older versions of qemu, so it is ok for me
>>> if it is removed (as long as that removal is not backported).
>>
>> Mmm, but if we have an active user who's testing them then they
>> probably shouldn't be in the frontline of boards to remove.
>> Which other boards do you test with mainline QEMU?
>>
> 
> For arm:
> 
> akita
> ast2500-evb
> beagle
> beaglexm
> borzoi
> collie
> cubieboard
> imx25-pdk
> integratorcp
> kzm
> mainstone
> midway
> mps2-an385
> overo
> palmetto-bmc
> raspi2
> realview-eb
> realview-eb-mpcore
> realview-pb-a8
> realview-pbx-a9
> romulus-bmc
> sabrelite
> smdkc210
> spitz
> terrier
> tosa
> versatileab
> versatilepb
> vexpress-a15
> vexpress-a15-a7
> vexpress-a9
> witherspoon-bmc
> xilinx-zynq-a9
> z2
> 
> Though not all of them are supported by upstream qemu. For some of them 
> I carry local patches,
> for others I use out-of-tree versions of qemu (beagle/beaglexm).

Are these patches upstream-able?

If not, what do we need to get them into upstream qemu?

Thanks,

Phil.

Re: [Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Guenter Roeck]

On Mon, Oct 29, 2018 at 04:03:40PM +0100, Philippe Mathieu-Daudé wrote:
> Hi Guenter,
> 
> On 29/10/18 15:09, Guenter Roeck wrote:
> >On 10/29/18 6:24 AM, Peter Maydell wrote:
> >>On 27 October 2018 at 12:04, Guenter Roeck <linux@roeck-us.net> wrote:
> >>>On 10/26/18 3:12 AM, Peter Maydell wrote:
> >>>>Hi Guenter; there's a proposal here to deprecate (and eventually
> >>>>remove) the 'collie' board (strongarm) from QEMU. Is that one of
> >>>>the ones you're currently using in your automated testing of Linux
> >>>>kernels on QEMU?
> >>>>
> >>>
> >>>Yes. I can run the test with older versions of qemu, so it is ok for me
> >>>if it is removed (as long as that removal is not backported).
> >>
> >>Mmm, but if we have an active user who's testing them then they
> >>probably shouldn't be in the frontline of boards to remove.
> >>Which other boards do you test with mainline QEMU?
> >>
> >
> >For arm:
> >
> >akita
> >ast2500-evb
> >beagle
> >beaglexm
> >borzoi
> >collie
> >cubieboard
> >imx25-pdk
> >integratorcp
> >kzm
> >mainstone
> >midway
> >mps2-an385
> >overo
> >palmetto-bmc
> >raspi2
> >realview-eb
> >realview-eb-mpcore
> >realview-pb-a8
> >realview-pbx-a9
> >romulus-bmc
> >sabrelite
> >smdkc210
> >spitz
> >terrier
> >tosa
> >versatileab
> >versatilepb
> >vexpress-a15
> >vexpress-a15-a7
> >vexpress-a9
> >witherspoon-bmc
> >xilinx-zynq-a9
> >z2
> >
> >Though not all of them are supported by upstream qemu. For some of them I
> >carry local patches,
> >for others I use out-of-tree versions of qemu (beagle/beaglexm).
> 
> Are these patches upstream-able?
> 

Some of the patches (eg to be able to boot a Linux kernel image for mps2-an385
directly from qemu, or zynq CPU clock rate changes to be able to do the same)
have been rejected. A few patches were submitted at some point but got lost.
I don't keep track, so I don't know the exact number. For some patches, such
as basic BCM283x CPRMAN support (needed to boot raspi2), a better
implementation was suggested, but didn't go anywhere as far as I know.

Some machines, such as beagle support, are from Linaro's tree and were never
upstreamed by Linaro. The Linaro branch is based off qemu 2.3, so applying
the changes to upstream qemu would be a major effort. I also use the m68k
branch from github.com:vivier/qemu-m68k.git for m68k tests.

I carry some 20+ patches locally in my qemu tree. Some may be obsolete or
not or no longer needed (my understanding of qemu is evolving). Sometimes,
if and when I find the time, I pick some and try to upstream, but I often
don't follow up if there is no response or if the requested changes are
too substantial.

Please feel free to have a look at https://github.com/groeck/qemu
(check the <version>-local branches) and let me know what might be
upstreamable. I'll be happy to (re-)submit the respective patches.
Note that I won't be able to make any substantial changes, though.
Time is a scarce commodity nowadays, unfortunately.

Guenter

Re: [Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Philippe Mathieu-Daudé]

On 29/10/18 21:24, Guenter Roeck wrote:
> On Mon, Oct 29, 2018 at 04:03:40PM +0100, Philippe Mathieu-Daudé wrote:
>> On 29/10/18 15:09, Guenter Roeck wrote:
>>> On 10/29/18 6:24 AM, Peter Maydell wrote:
>>>> On 27 October 2018 at 12:04, Guenter Roeck <linux@roeck-us.net> wrote:
>>>>> On 10/26/18 3:12 AM, Peter Maydell wrote:
>>>>>> Hi Guenter; there's a proposal here to deprecate (and eventually
>>>>>> remove) the 'collie' board (strongarm) from QEMU. Is that one of
>>>>>> the ones you're currently using in your automated testing of Linux
>>>>>> kernels on QEMU?
>>>>>>
>>>>>
>>>>> Yes. I can run the test with older versions of qemu, so it is ok for me
>>>>> if it is removed (as long as that removal is not backported).

[...]

>>> Though not all of them are supported by upstream qemu. For some of them I
>>> carry local patches,
>>> for others I use out-of-tree versions of qemu (beagle/beaglexm).
>>
>> Are these patches upstream-able?
>>
> 
> Some of the patches (eg to be able to boot a Linux kernel image for mps2-an385
> directly from qemu, or zynq CPU clock rate changes to be able to do the same)
> have been rejected. A few patches were submitted at some point but got lost.
> I don't keep track, so I don't know the exact number. For some patches, such
> as basic BCM283x CPRMAN support (needed to boot raspi2), a better
> implementation was suggested, but didn't go anywhere as far as I know.
> 
> Some machines, such as beagle support, are from Linaro's tree and were never
> upstreamed by Linaro. The Linaro branch is based off qemu 2.3, so applying
> the changes to upstream qemu would be a major effort. I also use the m68k
> branch from github.com:vivier/qemu-m68k.git for m68k tests.
> 
> I carry some 20+ patches locally in my qemu tree. Some may be obsolete or
> not or no longer needed (my understanding of qemu is evolving). Sometimes,
> if and when I find the time, I pick some and try to upstream, but I often
> don't follow up if there is no response or if the requested changes are
> too substantial.
> 
> Please feel free to have a look at https://github.com/groeck/qemu
> (check the <version>-local branches) and let me know what might be
> upstreamable. I'll be happy to (re-)submit the respective patches.
> Note that I won't be able to make any substantial changes, though.
> Time is a scarce commodity nowadays, unfortunately.

Thank you for pointing your work, I'll try to salvage what I can.

Regards,

Phil.

Re: [Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Markus Armbruster]

Guenter Roeck <linux@roeck-us.net> writes:

> On 10/29/18 6:24 AM, Peter Maydell wrote:
>> On 27 October 2018 at 12:04, Guenter Roeck <linux@roeck-us.net> wrote:
>>> On 10/26/18 3:12 AM, Peter Maydell wrote:
>>>> Hi Guenter; there's a proposal here to deprecate (and eventually
>>>> remove) the 'collie' board (strongarm) from QEMU. Is that one of
>>>> the ones you're currently using in your automated testing of Linux
>>>> kernels on QEMU?
>>>>
>>>
>>> Yes. I can run the test with older versions of qemu, so it is ok for me
>>> if it is removed (as long as that removal is not backported).
>>
>> Mmm, but if we have an active user who's testing them then they
>> probably shouldn't be in the frontline of boards to remove.
>> Which other boards do you test with mainline QEMU?
>>
>
> For arm:
>
> akita
> ast2500-evb
> beagle
> beaglexm
> borzoi
> collie
> cubieboard
> imx25-pdk
> integratorcp
> kzm
> mainstone
> midway
> mps2-an385
> overo
> palmetto-bmc
> raspi2
> realview-eb
> realview-eb-mpcore
> realview-pb-a8
> realview-pbx-a9
> romulus-bmc
> sabrelite
> smdkc210
> spitz
> terrier
> tosa
> versatileab
> versatilepb
> vexpress-a15
> vexpress-a15-a7
> vexpress-a9
> witherspoon-bmc
> xilinx-zynq-a9
> z2
>
> Though not all of them are supported by upstream qemu. For some of them I carry local patches,
> for others I use out-of-tree versions of qemu (beagle/beaglexm).

Would it be possible to get useful parts of your automated testing into
upstream QEMU's CI?

I'm asking because I think we should require at least a smoke test in CI
for all machine types, and drop the ones that lack it.  Not today, but
in the not-too-distant future.

Re: [Qemu-devel] [PATCH 0/2] Deprecate the "collie" machine and Strongarm devices

[Thomas Huth]

On 2018-10-29 13:24, Peter Maydell wrote:
> On 27 October 2018 at 12:04, Guenter Roeck <linux@roeck-us.net> wrote:
>> On 10/26/18 3:12 AM, Peter Maydell wrote:
>>> Hi Guenter; there's a proposal here to deprecate (and eventually
>>> remove) the 'collie' board (strongarm) from QEMU. Is that one of
>>> the ones you're currently using in your automated testing of Linux
>>> kernels on QEMU?
>>>
>>
>> Yes. I can run the test with older versions of qemu, so it is ok for me
>> if it is removed (as long as that removal is not backported).
> 
> Mmm, but if we have an active user who's testing them then they
> probably shouldn't be in the frontline of boards to remove.

Yes, I agree. I based my patch on the assumption that the board was
incomplete and not really usable (as mentioned in the patch
description), but if it is still usable to some degree, then please
disregard my patch.

 Thomas