[Qemu-devel] Buildbot Failures

[Qemu-devel] Buildbot Failures

[Charlie Shepherd]

Hi all,

This is an attempt to do some triaging of the 18 current buildbot 
failures[1].

default_ppc:
http://buildbot.b1-systems.de/qemu/builders/default_ppc/builds/729/steps/compile/logs/stdio
>    LINK  ppc-linux-user/qemu-ppc
> target-ppc/translate.o:(.text+0x53518): undefined reference to `tcg_conqt_i64'
> collect2: ld returned 1 exit status
> make[1]: *** [qemu-ppc] Error 1
> make: *** [subdir-ppc-linux-user] Error 2

This only failed last night, I'm not sure where the reference to 
tcg_conqt_i64 came from but it's a recent commit that could probably be 
bisected.



default_x86_64_rhel61:
http://buildbot.b1-systems.de/qemu/builders/default_x86_64_rhel61/builds/698/steps/test/logs/stdio
> GTESTER check-qtest-i386
> main-loop: WARNING: I/O thread spun for 1000 iterations
> **
> ERROR:tests/libqtest.c:69:init_socket: assertion failed (ret != -1): (-1 != -1)
This looks like a failure to initialise a socket in tests/libqtest.c, 
possibly due to /tmp/ not being accessible?



default_x86_64_rhel5:
http://buildbot.b1-systems.de/qemu/builders/default_x86_64_rhel5/builds/684/steps/compile/logs/stdio

>    File "/home/buildbot/slave-public/default_x86_64_rhel5/build/scripts/qapi-visit.py", line 23
>      full_name = name if not fn_prefix else "%s_%s" % (name, fn_prefix)
>                        ^
> SyntaxError: invalid syntax
> make: *** [qapi-visit.h] Error 1

This syntax was introduced in Python 2.5, patch to follow to convert 
this to valid Python 2.4 syntax.



default_openbsd_current (offline):
http://buildbot.b1-systems.de/qemu/builders/default_openbsd_current/builds/509/steps/compile/logs/stdio

>    AS    optionrom/multiboot.o
>    Building optionrom/multiboot.img
> ld: multiboot.o: relocation R_X86_64_16 can not be used when making a shared object; recompile with -fPIC
> multiboot.o: could not read symbols: Bad value
> gmake[1]: *** [multiboot.img] Error 1
> gmake: *** [romsubdir-optionrom] Error 2

This has been offline for a few weeks, I wonder if this issue has been 
fixed in master since?


default_openbsd_4.9:
http://buildbot.b1-systems.de/qemu/builders/default_openbsd_4.9/builds/700/steps/test/logs/stdio

> tests/test-aio.o(.text+0x3df): In function `test_source_timer_schedule':
> : undefined reference to `pipe2'
> tests/test-aio.o(.text+0x4baf): In function `test_timer_schedule':
> : undefined reference to `pipe2'
> collect2: ld returned 1 exit status
> gmake: *** [tests/test-aio] Error 1

Commit b53edf971 broke this, pipe2 is a Linux-specific function. Its 
availability can be tested for with CONFIG_PIPE2.


default_mingw32:
http://buildbot.b1-systems.de/qemu/builders/default_mingw32/builds/698/steps/test/logs/stdio

>    CC    tests/test-aio.o
> tests/test-aio.c: In function 'test_timer_schedule':
> tests/test-aio.c:378: warning: implicit declaration of function 'pipe2'
> tests/test-aio.c:378: warning: nested extern declaration of 'pipe2'
> tests/test-aio.c:378: error: 'O_NONBLOCK' undeclared (first use in this function)
> tests/test-aio.c:378: error: (Each undeclared identifier is reported only once
> tests/test-aio.c:378: error: for each function it appears in.)
> tests/test-aio.c:379: warning: implicit declaration of function 'aio_set_fd_handler'
> tests/test-aio.c:379: warning: nested extern declaration of 'aio_set_fd_handler'
> tests/test-aio.c:399: warning: implicit declaration of function 'sleep'
> tests/test-aio.c:399: warning: nested extern declaration of 'sleep'
> tests/test-aio.c: In function 'test_source_timer_schedule':
> tests/test-aio.c:719: error: 'O_NONBLOCK' undeclared (first use in this function)
> make: *** [tests/test-aio.o] Error 1

Commit b53edf971  broke this too - pipe2, O_NONBLOCK and sleep are all 
unavailable on MingGW. The test executables don't run from either 
checking the logs, but getting it compiling would be a good start.


default_i386_macosx (offline)
http://buildbot.b1-systems.de/qemu/builders/default_i386_macosx/builds/129/steps/git/logs/stdio
> fatal: Unable to look up git.qemu-project.org (port 9418) (Temporary failure in name resolution)

Is this builder in use? It's been offline for a long time.



block_openbsd_current:
block_openbsd_4.9:
block_mingw32:

The same errors as their non block counterparts.



trivial-patches_x86_64_debian_6_0:
http://buildbot.b1-systems.de/qemu/builders/trivial-patches_x86_64_debian_6_0/builds/717/steps/compile/logs/stdio
> block/ssh.o: In function `connect_to_ssh':
> ssh.c:(.text+0xf03): undefined reference to `libssh2_session_handshake'
> collect2: ld returned 1 exit status
> make[1]: Leaving directory `/home/build/qemu/trivial-patches_x86_64_debian_6_0/build'
> make[1]: *** [qemu-nbd] Error 1

QEMU requires libssh2-1.2.8 - the version this function was added in, so 
the buildslave needs updating to at least this libssh2 version. However 
this should have been caught at ./configure time, either block/ssh.o 
shouldn't be compiled because a suitable libssh2 isn't available or 
./configure should throw an error. There's a check in 
block/Makefile.objs so I'm not sure why this didn't happen, I guess it 
merits further investigation?



xen_x86_64_debian_6_0:
xen_i386_debian_6_0:
http://buildbot.b1-systems.de/qemu/builders/xen_i386_debian_6_0/builds/720/steps/configure/logs/stdio
> ERROR: unknown option --disable-debug-info
This builder is trying to build a 2 year old revision (!), so either 
needs to use a later revision or should be deactivated.




xen40:
xen41:
xen42:
xen_unstable:

> arm-softmmu/qemu-system-arm: error while loading shared libraries: libxenstore.so.3.0: cannot open shared object file: No such file or directory

libxenstore is a QEMU dependency, is it available on this buildslave?



Charlie

[1] http://buildbot.b1-systems.de/qemu/builders

[Qemu-devel] [PATCH] Remove Python 2.5 syntax from scripts/qapi-visit.py

[Charlie Shepherd]

On 29/08/2013 10:57, Charlie Shepherd wrote:

> default_x86_64_rhel5:
> http://buildbot.b1-systems.de/qemu/builders/default_x86_64_rhel5/builds/684/steps/compile/logs/stdio
>>     File "/home/buildbot/slave-public/default_x86_64_rhel5/build/scripts/qapi-visit.py", line 23
>>       full_name = name if not fn_prefix else "%s_%s" % (name, fn_prefix)
>>                         ^
>> SyntaxError: invalid syntax
>> make: *** [qapi-visit.h] Error 1
> This syntax was introduced in Python 2.5, patch to follow to convert
> this to valid Python 2.4 syntax.

The syntax `var = a if b else c` was added in Python 2.5, but QEMU has a

minimum Python version of 2.4, which chokes on this syntax. This patch

converts the new syntax to Python 2.4 compatible syntax.

---

  scripts/qapi-visit.py |   12 +++++++++---

  1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/scripts/qapi-visit.py b/scripts/qapi-visit.py

index 597cca4..5d04438 100644

--- a/scripts/qapi-visit.py

+++ b/scripts/qapi-visit.py

@@ -20,7 +20,10 @@ import errno

  def generate_visit_struct_fields(name, field_prefix, fn_prefix, members):

      substructs = []

      ret = ''

-    full_name = name if not fn_prefix else "%s_%s" % (name, fn_prefix)

+    if not fn_prefix:

+        full_name = name

+    else:

+        full_name = "%s_%s" % (name, fn_prefix)

  

      for argname, argentry, optional, structured in parse_args(members):

          if structured:

@@ -84,7 +87,10 @@ if (!error_is_set(errp)) {

  ''')

      push_indent()

  

-    full_name = name if not field_prefix else "%s_%s" % (field_prefix, name)

+    if not fn_prefix:

+        full_name = name

+    else:

+        full_name = "%s_%s" % (name, fn_prefix)

  

      if len(field_prefix):

          ret += mcgen('''

@@ -270,7 +276,7 @@ void visit_type_%(name)s(Visitor *m, %(name)s ** obj, const char *name, Error **

          if (!err) {

              switch ((*obj)->kind) {

  ''',

-                 name=name, type="type" if not discriminator else discriminator)

+                 name=name, type=(discriminator or "type"))

  

      for key in members:

          if not discriminator:

-- 

1.7.9.5

Re: [Qemu-devel] [PATCH] Remove Python 2.5 syntax from scripts/qapi-visit.py

[Stefan Hajnoczi]

On Thu, Aug 29, 2013 at 12:02 PM, Charlie Shepherd
<charlie@ctshepherd.com> wrote:
> On 29/08/2013 10:57, Charlie Shepherd wrote:
>
>> default_x86_64_rhel5:
>>
>> http://buildbot.b1-systems.de/qemu/builders/default_x86_64_rhel5/builds/684/steps/compile/logs/stdio
>>>
>>>     File
>>> "/home/buildbot/slave-public/default_x86_64_rhel5/build/scripts/qapi-visit.py",
>>> line 23
>>>       full_name = name if not fn_prefix else "%s_%s" % (name, fn_prefix)
>>>                         ^
>>> SyntaxError: invalid syntax
>>> make: *** [qapi-visit.h] Error 1
>>
>> This syntax was introduced in Python 2.5, patch to follow to convert
>> this to valid Python 2.4 syntax.
>
>
> The syntax `var = a if b else c` was added in Python 2.5, but QEMU has a
>
> minimum Python version of 2.4, which chokes on this syntax. This patch
>
> converts the new syntax to Python 2.4 compatible syntax.
>
> ---

Missing Signed-off-by: line.

Also it looks like the email is double-spaced.  Copy-paste or email
client issue?
> -    full_name = name if not field_prefix else "%s_%s" % (field_prefix,
> name)
>
> +    if not fn_prefix:
>
> +        full_name = name
>
> +    else:
>
> +        full_name = "%s_%s" % (name, fn_prefix)

Careful, (field_prefix, name) were swapped to (name, fn_prefix).

Re: [Qemu-devel] [PATCH] Remove Python 2.5 syntax from scripts/qapi-visit.py

[Charlie Shepherd]

On 29/08/2013 11:14, Stefan Hajnoczi wrote:
> On Thu, Aug 29, 2013 at 12:02 PM, Charlie Shepherd
> <charlie@ctshepherd.com> wrote:
>> On 29/08/2013 10:57, Charlie Shepherd wrote:
>>
>>> default_x86_64_rhel5:
>>>
>>> http://buildbot.b1-systems.de/qemu/builders/default_x86_64_rhel5/builds/684/steps/compile/logs/stdio
>>>>      File
>>>> "/home/buildbot/slave-public/default_x86_64_rhel5/build/scripts/qapi-visit.py",
>>>> line 23
>>>>        full_name = name if not fn_prefix else "%s_%s" % (name, fn_prefix)
>>>>                          ^
>>>> SyntaxError: invalid syntax
>>>> make: *** [qapi-visit.h] Error 1
>>> This syntax was introduced in Python 2.5, patch to follow to convert
>>> this to valid Python 2.4 syntax.
>>
>> The syntax `var = a if b else c` was added in Python 2.5, but QEMU has a
>>
>> minimum Python version of 2.4, which chokes on this syntax. This patch
>>
>> converts the new syntax to Python 2.4 compatible syntax.
>>
>> ---
> Missing Signed-off-by: line.
>
> Also it looks like the email is double-spaced.  Copy-paste or email
> client issue?

Thunderbird strikes again. I'll send a respin with git send-email.

>> -    full_name = name if not field_prefix else "%s_%s" % (field_prefix,
>> name)
>>
>> +    if not fn_prefix:
>>
>> +        full_name = name
>>
>> +    else:
>>
>> +        full_name = "%s_%s" % (name, fn_prefix)
> Careful, (field_prefix, name) were swapped to (name, fn_prefix).

Good catch.


Charlie

[Qemu-devel] [PATCH v2] Remove Python 2.5 syntax from scripts/qapi-visit.py

[Charlie Shepherd]

The syntax `var = a if b else c` was added in Python 2.5, but QEMU has a
minimum Python version of 2.4, which chokes on this syntax. This patch
converts the new syntax to Python 2.4 compatible syntax.

Signed-off-by: Charlie Shepherd <charlie@ctshepherd.com>
---
 scripts/qapi-visit.py |   12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/scripts/qapi-visit.py b/scripts/qapi-visit.py
index 597cca4..cdf40c9 100644
--- a/scripts/qapi-visit.py
+++ b/scripts/qapi-visit.py
@@ -20,7 +20,10 @@ import errno
 def generate_visit_struct_fields(name, field_prefix, fn_prefix, members):
     substructs = []
     ret = ''
-    full_name = name if not fn_prefix else "%s_%s" % (name, fn_prefix)
+    if not fn_prefix:
+        full_name = name
+    else:
+        full_name = "%s_%s" % (name, fn_prefix)
 
     for argname, argentry, optional, structured in parse_args(members):
         if structured:
@@ -84,7 +87,10 @@ if (!error_is_set(errp)) {
 ''')
     push_indent()
 
-    full_name = name if not field_prefix else "%s_%s" % (field_prefix, name)
+    if not field_prefix:
+        full_name = name
+    else:
+        full_name = "%s_%s" % (field_prefix, name)
 
     if len(field_prefix):
         ret += mcgen('''
@@ -270,7 +276,7 @@ void visit_type_%(name)s(Visitor *m, %(name)s ** obj, const char *name, Error **
         if (!err) {
             switch ((*obj)->kind) {
 ''',
-                 name=name, type="type" if not discriminator else discriminator)
+                 name=name, type=(discriminator or "type"))
 
     for key in members:
         if not discriminator:
-- 
1.7.9.5

Re: [Qemu-devel] [PATCH v2] Remove Python 2.5 syntax from scripts/qapi-visit.py

[Stefan Hajnoczi]

On Thu, Aug 29, 2013 at 1:03 PM, Charlie Shepherd
<charlie@ctshepherd.com> wrote:
> The syntax `var = a if b else c` was added in Python 2.5, but QEMU has a
> minimum Python version of 2.4, which chokes on this syntax. This patch
> converts the new syntax to Python 2.4 compatible syntax.
>
> Signed-off-by: Charlie Shepherd <charlie@ctshepherd.com>
> ---

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>

Re: [Qemu-devel] Buildbot Failures

[Anthony PERARD]

Hi,

On 29/08/13 10:57, Charlie Shepherd wrote:
> xen40:
> xen41:
> xen42:
> xen_unstable:
> 
>> > arm-softmmu/qemu-system-arm: error while loading shared libraries: libxenstore.so.3.0: cannot open shared object file: No such file or directory
> libxenstore is a QEMU dependency, is it available on this buildslave?

Yes, but not system wide.

So, we would need to provide LD_LIBRARY_PATH to the tests. And those are
the paths for the different builders:

xen40:
/home/build/install/xen-4.0.x/lib/

xen41:
/home/build/install/xen-4.1.x/lib/

xen42:
/home/build/install/xen-4.2.x/lib/

xen_unstable:
/home/build/install/xen-unstable/lib/

Thanks for locking at this.

-- 
Anthony PERARD

Re: [Qemu-devel] Buildbot Failures

[Stefan Hajnoczi]

On Thu, Aug 29, 2013 at 12:19 PM, Anthony PERARD
<anthony.perard@citrix.com> wrote:
> On 29/08/13 10:57, Charlie Shepherd wrote:
>> xen40:
>> xen41:
>> xen42:
>> xen_unstable:
>>
>>> > arm-softmmu/qemu-system-arm: error while loading shared libraries: libxenstore.so.3.0: cannot open shared object file: No such file or directory
>> libxenstore is a QEMU dependency, is it available on this buildslave?
>
> Yes, but not system wide.
>
> So, we would need to provide LD_LIBRARY_PATH to the tests. And those are
> the paths for the different builders:
>
> xen40:
> /home/build/install/xen-4.0.x/lib/
>
> xen41:
> /home/build/install/xen-4.1.x/lib/
>
> xen42:
> /home/build/install/xen-4.2.x/lib/
>
> xen_unstable:
> /home/build/install/xen-unstable/lib/

The build configuration is here:

https://github.com/b1-systems/buildbot/blob/master/qemu-master.cfg

In particular, look at create_build_factory().

You can fix this and then raise a pull request on the GitHub
repository.  I guess you just need to customize the if run_tests:
stanza to also include an optional LD_LIBRARY_PATH.

Buildbot Python API documentation is here:
http://docs.buildbot.net/current/index.html

Stefan

Re: [Qemu-devel] Buildbot Failures

[Charlie Shepherd]

On 29/08/2013 11:23, Stefan Hajnoczi wrote:
> On Thu, Aug 29, 2013 at 12:19 PM, Anthony PERARD
> <anthony.perard@citrix.com> wrote:
>> On 29/08/13 10:57, Charlie Shepherd wrote:
>>> xen40:
>>> xen41:
>>> xen42:
>>> xen_unstable:
>>>
>>>>> arm-softmmu/qemu-system-arm: error while loading shared libraries: libxenstore.so.3.0: cannot open shared object file: No such file or directory
>>> libxenstore is a QEMU dependency, is it available on this buildslave?
>> Yes, but not system wide.
>>
>> So, we would need to provide LD_LIBRARY_PATH to the tests. And those are
>> the paths for the different builders:
>>
>> xen40:
>> /home/build/install/xen-4.0.x/lib/
>>
>> xen41:
>> /home/build/install/xen-4.1.x/lib/
>>
>> xen42:
>> /home/build/install/xen-4.2.x/lib/
>>
>> xen_unstable:
>> /home/build/install/xen-unstable/lib/
> The build configuration is here:
>
> https://github.com/b1-systems/buildbot/blob/master/qemu-master.cfg
>
> In particular, look at create_build_factory().
>
> You can fix this and then raise a pull request on the GitHub
> repository.  I guess you just need to customize the if run_tests:
> stanza to also include an optional LD_LIBRARY_PATH.

https://github.com/b1-systems/buildbot/pull/7


Charlie

Re: [Qemu-devel] Buildbot Failures

[Stefan Hajnoczi]

On Thu, Aug 29, 2013 at 11:57 AM, Charlie Shepherd
<charlie@ctshepherd.com> wrote:
> This is an attempt to do some triaging of the 18 current buildbot
> failures[1].

Thanks for doing this!

> trivial-patches_x86_64_debian_6_0:
> http://buildbot.b1-systems.de/qemu/builders/trivial-patches_x86_64_debian_6_0/builds/717/steps/compile/logs/stdio
>>
>> block/ssh.o: In function `connect_to_ssh':
>> ssh.c:(.text+0xf03): undefined reference to `libssh2_session_handshake'
>> collect2: ld returned 1 exit status
>> make[1]: Leaving directory
>> `/home/build/qemu/trivial-patches_x86_64_debian_6_0/build'
>> make[1]: *** [qemu-nbd] Error 1
>
>
> QEMU requires libssh2-1.2.8 - the version this function was added in, so the
> buildslave needs updating to at least this libssh2 version. However this
> should have been caught at ./configure time, either block/ssh.o shouldn't be
> compiled because a suitable libssh2 isn't available or ./configure should
> throw an error. There's a check in block/Makefile.objs so I'm not sure why
> this didn't happen, I guess it merits further investigation?

This builder is using my old trivial-patches repository.  Michael
Tokarev has taken over trivial-patches maintainership.  I have raised
a pull request to drop my stale repository from the buildbot:
https://github.com/b1-systems/buildbot/pull/6

Stefan

Re: [Qemu-devel] Buildbot Failures

[Alex Bligh]

> 
> default_x86_64_rhel61:
> http://buildbot.b1-systems.de/qemu/builders/default_x86_64_rhel61/builds/698/steps/test/logs/stdio
>> GTESTER check-qtest-i386
>> main-loop: WARNING: I/O thread spun for 1000 iterations
>> **
>> ERROR:tests/libqtest.c:69:init_socket: assertion failed (ret != -1): (-1 != -1)
> This looks like a failure to initialise a socket in tests/libqtest.c, possibly due to /tmp/ not being accessible?

This happens when a failed test run has not cleared
up its sockets, and the same PID is reused. Better error
trapping within the tester would be useful. Alternatively
the buildbot could clear temporary files out before running.

> default_openbsd_4.9:
> http://buildbot.b1-systems.de/qemu/builders/default_openbsd_4.9/builds/700/steps/test/logs/stdio
> 
>> tests/test-aio.o(.text+0x3df): In function `test_source_timer_schedule':
>> : undefined reference to `pipe2'
>> tests/test-aio.o(.text+0x4baf): In function `test_timer_schedule':
>> : undefined reference to `pipe2'
>> collect2: ld returned 1 exit status
>> gmake: *** [tests/test-aio] Error 1
> 
> Commit b53edf971 broke this, pipe2 is a Linux-specific function. Its availability can be tested for with CONFIG_PIPE2.

Guessing that's mine. I'll take a look. A simple fix might
be to only run the test on Linux for the time being.

-- 
Alex Bligh

Re: [Qemu-devel] Buildbot Failures

[Brad Smith]

On 29/08/13 5:57 AM, Charlie Shepherd wrote:
> Hi all,
>
> This is an attempt to do some triaging of the 18 current buildbot
> failures[1].
>
> default_openbsd_current (offline):
> http://buildbot.b1-systems.de/qemu/builders/default_openbsd_current/builds/509/steps/compile/logs/stdio
>
>
>>    AS    optionrom/multiboot.o
>>    Building optionrom/multiboot.img
>> ld: multiboot.o: relocation R_X86_64_16 can not be used when making a
>> shared object; recompile with -fPIC
>> multiboot.o: could not read symbols: Bad value
>> gmake[1]: *** [multiboot.img] Error 1
>> gmake: *** [romsubdir-optionrom] Error 2
>
> This has been offline for a few weeks, I wonder if this issue has been
> fixed in master since?

I have been meaning to try and come up with a fix for this issue for 
awhile now. This has come about because OpenBSD 5.3 and newer now 
utilizes PIE by default on most of our architectures and building the 
ROM images needs to have PIE disabled. We have a patch in our ports tree 
to deal with this but it is not appropriate for upstream as is.

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.