Re: [Qemu-devel] [PULL] Memory API conversions, batch 5

[Blue Swirl]

On Thu, Aug 25, 2011 at 7:02 PM, Anthony Liguori <address@hidden> wrote:
> On 08/25/2011 01:58 PM, Blue Swirl wrote:
>>
>> On Thu, Aug 25, 2011 at 6:27 PM, Edgar E. Iglesias
>> <address@hidden>  wrote:
>>>
>>> On Thu, Aug 25, 2011 at 08:04:34AM -0500, Anthony Liguori wrote:
>>>>
>>>> On 08/25/2011 03:01 AM, Avi Kivity wrote:
>>>>>
>>>>> Please pull from
>>>>>
>>>>> git://git.kernel.org/pub/scm/virt/kvm/qemu-kvm.git memory/batch
>>>>>
>>>>> to receive the fifth memory API conversions batch. There is really
>>>>> nothing notable, except perhaps the first use of the transaction API.
>>>>>
>>>>> While the patches have been on the list for just a short time, they
>>>>> have
>>>>> received quite extensive review, for which I am thankful.
>>>>
>>>> Pulled.  Thanks.
>>>
>>> Guys, please start testing before applying these patches. It's not hard,
>>> there are images on the wiki with scripts that area easy to run.
>>
>> It's also possible to revert offending patches. But a better solution
>> would be to have a comprehensive test setup to do builds and run
>> checks on selected trees. Before a pull to main tree, the test results
>> should be available.
>
> Before a pull *request*.  Semantically, pull requests should be treated as a
> push.  It's the submaintainers responsibility to test the bits they do a
> pull request for.

Then all maintainers should have the same understanding of what to test.

> That said, the logistics of testing a series like this is mind numbingly
> complex.  We really need to all work on better automation for this.  In the
> interim, I think we're just going to have to leave with occasional breakage
> from changes like this and make sure it doesn't happen too frequently.

Alex's simple script idea for kernel devs could be reused here. The
following would already catch a few errors:
#!/bin/sh
set -e
mkdir obj
cd obj
../configure
make -j
for f in ./*-softmmu/qemu-system-* ./i386-softmmu/qemu; do
  echo running $f, exit with Ctrl-a x:
  $f
done

This could be integrated to Makefile as well, maybe even generate
timestamp files which would be read by git commit hooks to add extra
lines about test status.

> Regards,
>
> Anthony Liguori
>
>>
>>>
>>> Cheers
>>>
>>> $ gdb --args
>>> ~/src/c/qemu/build-qemu/microblaze-softmmu/qemu-system-microblaze -kernel
>>> linux.bin -serial stdio
>>> GNU gdb (Ubuntu/Linaro 7.3-0ubuntu1) 7.3-2011.08~bzr
>>> Copyright (C) 2011 Free Software Foundation, Inc.
>>> License GPLv3+: GNU GPL version 3 or
>>> later<http://gnu.org/licenses/gpl.html>
>>> This is free software: you are free to change and redistribute it.
>>> There is NO WARRANTY, to the extent permitted by law.  Type "show
>>> copying"
>>> and "show warranty" for details.
>>> This GDB was configured as "x86_64-linux-gnu".
>>> For bug reporting instructions, please see:
>>> <http://www.gnu.org/software/gdb/bugs/>...
>>> Reading symbols from
>>> /home/edgar/src/c/qemu/build-qemu/microblaze-softmmu/qemu-system-microblaze...done.
>>> (gdb) run
>>> Starting program:
>>> /home/edgar/src/c/qemu/build-qemu/microblaze-softmmu/qemu-system-microblaze
>>> -kernel linux.bin -serial stdio
>>> [Thread debugging using libthread_db enabled]
>>> [New Thread 0x7fffeef3e700 (LWP 25891)]
>>>
>>> Program received signal SIGSEGV, Segmentation fault.
>>> 0x000000000041be54 in memory_region_init_rom_device (mr=0xfb6900,
>>>    ops=<optimized out>, dev=<optimized out>,
>>>    name=0x564c5d "petalogix_s3adsp1800.flash", size=16777216)
>>>    at /home/edgar/src/c/qemu/qemu/memory.c:973
>>> 973         mr->ram_addr |=
>>> cpu_register_io_memory(memory_region_read_thunk,
>>> (gdb)
>>>
>>>
>>
>>
>
>