[PATCH obv/pushed] MIPS ptrace build fixes

[PATCH obv/pushed] MIPS ptrace build fixes

[Simon Marchi]

FYI, I am pushing this as obvious.

Since Pedro's ptrace cleanups, the MIPS buildbot compilation fails.
Code in MIPS native uses ptrace with 3 arguments, where ptrace requires
4.  When looking at the definition of ptrace in
/usr/include/sys/ptrace.h, it shows that it takes a variable number of
arguments.  The wrapper macro in nat/gdb_ptrace.h takes a fixed number
of arguments (4).  That would explain why it used to work and stopped.

I am pushing this as obvious, tell me if there is any problem.

I built-tested this with a MIPS toolchain (ct-ng), but I don't have any
setup to test it.  At least it should put back the buildbot builder in a
better shape.

gdb/ChangeLog:

	* mips-linux-nat.c (write_watchpoint_regs): Add NULL as ptrace's 4th
	parameter.
	(mips_linux_new_thread): Likewise.
	* nat/mips-linux-watch.c (mips_linux_read_watch_registers): Likewise.

gdb/gdbserver/ChangeLog:

	* linux-mips-low.c (mips_linux_prepare_to_resume): Add NULL as
	ptrace's 4th parameter.
---
 gdb/ChangeLog                  | 7 +++++++
 gdb/gdbserver/ChangeLog        | 5 +++++
 gdb/gdbserver/linux-mips-low.c | 2 +-
 gdb/mips-linux-nat.c           | 4 ++--
 gdb/nat/mips-linux-watch.c     | 2 +-
 5 files changed, 16 insertions(+), 4 deletions(-)

diff --git a/gdb/ChangeLog b/gdb/ChangeLog
index 0c24f0c..8dee2da 100644
--- a/gdb/ChangeLog
+++ b/gdb/ChangeLog
@@ -1,3 +1,10 @@
+2015-07-29  Simon Marchi  <simon.marchi@ericsson.com>
+
+	* mips-linux-nat.c (write_watchpoint_regs): Add NULL as ptrace's 4th
+	parameter.
+	(mips_linux_new_thread): Likewise.
+	* nat/mips-linux-watch.c (mips_linux_read_watch_registers): Likewise.
+
 2015-07-29  Patrick Palka  <patrick@parcs.ath.cx>
 
 	* top.c: Include "tui/tui.h".
diff --git a/gdb/gdbserver/ChangeLog b/gdb/gdbserver/ChangeLog
index 5b4236a..2e039b5 100644
--- a/gdb/gdbserver/ChangeLog
+++ b/gdb/gdbserver/ChangeLog
@@ -1,3 +1,8 @@
+2015-07-29  Simon Marchi  <simon.marchi@ericsson.com>
+
+	* linux-mips-low.c (mips_linux_prepare_to_resume): Add NULL as
+	ptrace's 4th parameter.
+
 2015-07-27  Yao Qi  <yao.qi@linaro.org>
 
 	* configure.srv (case aarch64*-*-linux*): Don't set
diff --git a/gdb/gdbserver/linux-mips-low.c b/gdb/gdbserver/linux-mips-low.c
index d3b01d6..770f0df 100644
--- a/gdb/gdbserver/linux-mips-low.c
+++ b/gdb/gdbserver/linux-mips-low.c
@@ -427,7 +427,7 @@ mips_linux_prepare_to_resume (struct lwp_info *lwp)
 	  int tid = ptid_get_lwp (ptid);
 
 	  if (-1 == ptrace (PTRACE_SET_WATCH_REGS, tid,
-			    &priv->watch_mirror))
+			    &priv->watch_mirror, NULL))
 	    perror_with_name ("Couldn't write watch register");
 	}
 
diff --git a/gdb/mips-linux-nat.c b/gdb/mips-linux-nat.c
index 961cb6f..6df618d 100644
--- a/gdb/mips-linux-nat.c
+++ b/gdb/mips-linux-nat.c
@@ -614,7 +614,7 @@ write_watchpoint_regs (void)
   ALL_LWPS (lp)
     {
       tid = ptid_get_lwp (lp->ptid);
-      if (ptrace (PTRACE_SET_WATCH_REGS, tid, &watch_mirror) == -1)
+      if (ptrace (PTRACE_SET_WATCH_REGS, tid, &watch_mirror, NULL) == -1)
 	perror_with_name (_("Couldn't write debug register"));
     }
   return 0;
@@ -634,7 +634,7 @@ mips_linux_new_thread (struct lwp_info *lp)
     return;
 
   tid = ptid_get_lwp (lp->ptid);
-  if (ptrace (PTRACE_SET_WATCH_REGS, tid, &watch_mirror) == -1)
+  if (ptrace (PTRACE_SET_WATCH_REGS, tid, &watch_mirror, NULL) == -1)
     perror_with_name (_("Couldn't write debug register"));
 }
 
diff --git a/gdb/nat/mips-linux-watch.c b/gdb/nat/mips-linux-watch.c
index 02d83f6..231dbe0 100644
--- a/gdb/nat/mips-linux-watch.c
+++ b/gdb/nat/mips-linux-watch.c
@@ -164,7 +164,7 @@ mips_linux_read_watch_registers (long lwpid,
 {
   if (force || *watch_readback_valid == 0)
     {
-      if (ptrace (PTRACE_GET_WATCH_REGS, lwpid, watch_readback) == -1)
+      if (ptrace (PTRACE_GET_WATCH_REGS, lwpid, watch_readback, NULL) == -1)
 	{
 	  *watch_readback_valid = -1;
 	  return 0;
-- 
2.1.4

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Simon Marchi]

On 15-07-29 05:19 PM, Simon Marchi wrote:
> FYI, I am pushing this as obvious.
> 
> Since Pedro's ptrace cleanups, the MIPS buildbot compilation fails.
> Code in MIPS native uses ptrace with 3 arguments, where ptrace requires
> 4.  When looking at the definition of ptrace in
> /usr/include/sys/ptrace.h, it shows that it takes a variable number of
> arguments.  The wrapper macro in nat/gdb_ptrace.h takes a fixed number
> of arguments (4).  That would explain why it used to work and stopped.
> 
> I am pushing this as obvious, tell me if there is any problem.
> 
> I built-tested this with a MIPS toolchain (ct-ng), but I don't have any
> setup to test it.  At least it should put back the buildbot builder in a
> better shape.
> 
> gdb/ChangeLog:
> 
> 	* mips-linux-nat.c (write_watchpoint_regs): Add NULL as ptrace's 4th
> 	parameter.
> 	(mips_linux_new_thread): Likewise.
> 	* nat/mips-linux-watch.c (mips_linux_read_watch_registers): Likewise.
> 
> gdb/gdbserver/ChangeLog:
> 
> 	* linux-mips-low.c (mips_linux_prepare_to_resume): Add NULL as
> 	ptrace's 4th parameter.
> ---
>  gdb/ChangeLog                  | 7 +++++++
>  gdb/gdbserver/ChangeLog        | 5 +++++
>  gdb/gdbserver/linux-mips-low.c | 2 +-
>  gdb/mips-linux-nat.c           | 4 ++--
>  gdb/nat/mips-linux-watch.c     | 2 +-
>  5 files changed, 16 insertions(+), 4 deletions(-)
> 
> diff --git a/gdb/ChangeLog b/gdb/ChangeLog
> index 0c24f0c..8dee2da 100644
> --- a/gdb/ChangeLog
> +++ b/gdb/ChangeLog
> @@ -1,3 +1,10 @@
> +2015-07-29  Simon Marchi  <simon.marchi@ericsson.com>
> +
> +	* mips-linux-nat.c (write_watchpoint_regs): Add NULL as ptrace's 4th
> +	parameter.
> +	(mips_linux_new_thread): Likewise.
> +	* nat/mips-linux-watch.c (mips_linux_read_watch_registers): Likewise.
> +
>  2015-07-29  Patrick Palka  <patrick@parcs.ath.cx>
>  
>  	* top.c: Include "tui/tui.h".
> diff --git a/gdb/gdbserver/ChangeLog b/gdb/gdbserver/ChangeLog
> index 5b4236a..2e039b5 100644
> --- a/gdb/gdbserver/ChangeLog
> +++ b/gdb/gdbserver/ChangeLog
> @@ -1,3 +1,8 @@
> +2015-07-29  Simon Marchi  <simon.marchi@ericsson.com>
> +
> +	* linux-mips-low.c (mips_linux_prepare_to_resume): Add NULL as
> +	ptrace's 4th parameter.
> +
>  2015-07-27  Yao Qi  <yao.qi@linaro.org>
>  
>  	* configure.srv (case aarch64*-*-linux*): Don't set
> diff --git a/gdb/gdbserver/linux-mips-low.c b/gdb/gdbserver/linux-mips-low.c
> index d3b01d6..770f0df 100644
> --- a/gdb/gdbserver/linux-mips-low.c
> +++ b/gdb/gdbserver/linux-mips-low.c
> @@ -427,7 +427,7 @@ mips_linux_prepare_to_resume (struct lwp_info *lwp)
>  	  int tid = ptid_get_lwp (ptid);
>  
>  	  if (-1 == ptrace (PTRACE_SET_WATCH_REGS, tid,
> -			    &priv->watch_mirror))
> +			    &priv->watch_mirror, NULL))
>  	    perror_with_name ("Couldn't write watch register");
>  	}
>  
> diff --git a/gdb/mips-linux-nat.c b/gdb/mips-linux-nat.c
> index 961cb6f..6df618d 100644
> --- a/gdb/mips-linux-nat.c
> +++ b/gdb/mips-linux-nat.c
> @@ -614,7 +614,7 @@ write_watchpoint_regs (void)
>    ALL_LWPS (lp)
>      {
>        tid = ptid_get_lwp (lp->ptid);
> -      if (ptrace (PTRACE_SET_WATCH_REGS, tid, &watch_mirror) == -1)
> +      if (ptrace (PTRACE_SET_WATCH_REGS, tid, &watch_mirror, NULL) == -1)
>  	perror_with_name (_("Couldn't write debug register"));
>      }
>    return 0;
> @@ -634,7 +634,7 @@ mips_linux_new_thread (struct lwp_info *lp)
>      return;
>  
>    tid = ptid_get_lwp (lp->ptid);
> -  if (ptrace (PTRACE_SET_WATCH_REGS, tid, &watch_mirror) == -1)
> +  if (ptrace (PTRACE_SET_WATCH_REGS, tid, &watch_mirror, NULL) == -1)
>      perror_with_name (_("Couldn't write debug register"));
>  }
>  
> diff --git a/gdb/nat/mips-linux-watch.c b/gdb/nat/mips-linux-watch.c
> index 02d83f6..231dbe0 100644
> --- a/gdb/nat/mips-linux-watch.c
> +++ b/gdb/nat/mips-linux-watch.c
> @@ -164,7 +164,7 @@ mips_linux_read_watch_registers (long lwpid,
>  {
>    if (force || *watch_readback_valid == 0)
>      {
> -      if (ptrace (PTRACE_GET_WATCH_REGS, lwpid, watch_readback) == -1)
> +      if (ptrace (PTRACE_GET_WATCH_REGS, lwpid, watch_readback, NULL) == -1)
>  	{
>  	  *watch_readback_valid = -1;
>  	  return 0;

Also, I cancelled the current build as well as all pending builds before this commit
on the MIPS builder, in case you were wondering where they all went.  There is no
point in spending over 24 hours building code we know won't build.

Simon

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Pedro Alves]

On 07/29/2015 10:26 PM, Simon Marchi wrote:
> On 15-07-29 05:19 PM, Simon Marchi wrote:
>> FYI, I am pushing this as obvious.
>>
>> Since Pedro's ptrace cleanups, the MIPS buildbot compilation fails.

In my defense, I did get a build fail email report, but it points at:

 http://gdb-build.sergiodj.net/builders/Debian-MIPS-m64/builds/77

and looking at:

 http://gdb-build.sergiodj.net/builders/Debian-MIPS-m64/builds/77/steps/compile%20gdb/logs/stdio

it shows the build had failed for an unrelated issue:

 tdep.Tpo ../../binutils-gdb/gdb/i386obsd-tdep.c
 In file included from ../../binutils-gdb/gdb/common/common-defs.h:32:0,
                  from ../../binutils-gdb/gdb/defs.h:28,
                  from ../../binutils-gdb/gdb/i386obsd-tdep.c:20:
 build-gnulib/import/stdlib.h:41:20: fatal error: /usr/lib/gcc/mips-linux-gnu/4.9/include/ctddef.h: No such file or directory
  #include <stddef.h>
                     ^
 compilation terminated.
 The bug is not reproducible, so it is likely a hardware or OS problem.
 make[2]: *** [i386obsd-tdep.o] Error 1
 Makefile:1133: recipe for target 'i386obsd-tdep.o' failed

So I just ignored it.  Guess I was "lucky".

>> Code in MIPS native uses ptrace with 3 arguments, where ptrace requires
>> 4.  When looking at the definition of ptrace in
>> /usr/include/sys/ptrace.h, it shows that it takes a variable number of
>> arguments.  The wrapper macro in nat/gdb_ptrace.h takes a fixed number
>> of arguments (4).  That would explain why it used to work and stopped.
>>
>> I am pushing this as obvious, tell me if there is any problem.

Yes, that's the right fix:
 https://sourceware.org/ml/gdb-patches/2015-07/msg00721.html

> Also, I cancelled the current build as well as all pending builds before this commit
> on the MIPS builder, in case you were wondering where they all went.  There is no
> point in spending over 24 hours building code we know won't build.

Thanks.

-- 
Pedro Alves

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Sergio Durigan Junior]

On Wednesday, July 29 2015, Pedro Alves wrote:

> On 07/29/2015 10:26 PM, Simon Marchi wrote:
>> On 15-07-29 05:19 PM, Simon Marchi wrote:
>>> FYI, I am pushing this as obvious.
>>>
>>> Since Pedro's ptrace cleanups, the MIPS buildbot compilation fails.
>
> In my defense, I did get a build fail email report, but it points at:
>
>  http://gdb-build.sergiodj.net/builders/Debian-MIPS-m64/builds/77
>
> and looking at:
>
>  http://gdb-build.sergiodj.net/builders/Debian-MIPS-m64/builds/77/steps/compile%20gdb/logs/stdio
>
> it shows the build had failed for an unrelated issue:
>
>  tdep.Tpo ../../binutils-gdb/gdb/i386obsd-tdep.c
>  In file included from ../../binutils-gdb/gdb/common/common-defs.h:32:0,
>                   from ../../binutils-gdb/gdb/defs.h:28,
>                   from ../../binutils-gdb/gdb/i386obsd-tdep.c:20:
>  build-gnulib/import/stdlib.h:41:20: fatal error: /usr/lib/gcc/mips-linux-gnu/4.9/include/ctddef.h: No such file or directory
>   #include <stddef.h>
>                      ^
>  compilation terminated.
>  The bug is not reproducible, so it is likely a hardware or OS problem.
>  make[2]: *** [i386obsd-tdep.o] Error 1
>  Makefile:1133: recipe for target 'i386obsd-tdep.o' failed
>
> So I just ignored it.  Guess I was "lucky".

FWIW, I've been noticing random errors in the MIPS buildslave.  I even
spotted an ICE once (I think Alan Modra reported it to me).  Not sure if
it is because of the load or something else.

Thanks,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Peter Schauer]

> On Wednesday, July 29 2015, Pedro Alves wrote:
> 
> > On 07/29/2015 10:26 PM, Simon Marchi wrote:
> >> On 15-07-29 05:19 PM, Simon Marchi wrote:
> >>> FYI, I am pushing this as obvious.
> >>>
> >>> Since Pedro's ptrace cleanups, the MIPS buildbot compilation fails.
> >
> > In my defense, I did get a build fail email report, but it points at:
> >
> >  http://gdb-build.sergiodj.net/builders/Debian-MIPS-m64/builds/77
> >
> > and looking at:
> >
> >  http://gdb-build.sergiodj.net/builders/Debian-MIPS-m64/builds/77/steps/compile%20gdb/logs/stdio
> >
> > it shows the build had failed for an unrelated issue:
> >
> >  tdep.Tpo ../../binutils-gdb/gdb/i386obsd-tdep.c
> >  In file included from ../../binutils-gdb/gdb/common/common-defs.h:32:0,
> >                   from ../../binutils-gdb/gdb/defs.h:28,
> >                   from ../../binutils-gdb/gdb/i386obsd-tdep.c:20:
> >  build-gnulib/import/stdlib.h:41:20: fatal error: /usr/lib/gcc/mips-linux-gnu/4.9/include/ctddef.h: No such file or directory
> >   #include <stddef.h>
> >                      ^

This looks like a memory problem on the build slave, it seems that
memory flipped from 0x73 ('s') to 0x63 ('c').

> >  compilation terminated.
> >  The bug is not reproducible, so it is likely a hardware or OS problem.
> >  make[2]: *** [i386obsd-tdep.o] Error 1
> >  Makefile:1133: recipe for target 'i386obsd-tdep.o' failed
> >
> > So I just ignored it.  Guess I was "lucky".
> 
> FWIW, I've been noticing random errors in the MIPS buildslave.  I even
> spotted an ICE once (I think Alan Modra reported it to me).  Not sure if
> it is because of the load or something else.

So these might have been caused by memory problems too.

-- 
Peter Schauer			Peter.Schauer@mytum.de

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Pedro Alves]

Hi guys,

The MIPS build slave continues showing random build failures,
which results in spurious and confusing "you broke gdb" nag emails.

We should at least disable email reports for this build
slave until this is sorted out.

On 07/30/2015 10:13 AM, Peter Schauer wrote:
>> On Wednesday, July 29 2015, Pedro Alves wrote:
>>>
>>>  http://gdb-build.sergiodj.net/builders/Debian-MIPS-m64/builds/77/steps/compile%20gdb/logs/stdio
>>>
>>> it shows the build had failed for an unrelated issue:
>>>
> This looks like a memory problem on the build slave, it seems that
> memory flipped from 0x73 ('s') to 0x63 ('c').

OOC, how did you figure this out?

Thanks,
Pedro Alves

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Peter Schauer]

> 
> Hi guys,
> 
> The MIPS build slave continues showing random build failures,
> which results in spurious and confusing "you broke gdb" nag emails.
> 
> We should at least disable email reports for this build
> slave until this is sorted out.
> 
> On 07/30/2015 10:13 AM, Peter Schauer wrote:
> >> On Wednesday, July 29 2015, Pedro Alves wrote:
> >>>
> >>>  http://gdb-build.sergiodj.net/builders/Debian-MIPS-m64/builds/77/steps/compile%20gdb/logs/stdio
> >>>
> >>> it shows the build had failed for an unrelated issue:
> >>>
> > This looks like a memory problem on the build slave, it seems that
> > memory flipped from 0x73 ('s') to 0x63 ('c').
> 
> OOC, how did you figure this out?

From these lines:

build-gnulib/import/stdlib.h:41:20: fatal error: /usr/lib/gcc/mips-linux-gnu/4.9/include/ctddef.h: No such file or directory
  #include <stddef.h>
                     ^

gcc lists the correct source include (stddef.h), but then complains
about not finding ctddef.h, so it looks like memory flipped somewhere
between (or while) reading the include file name and then trying to
access the include file.

-- 
Peter Schauer			Peter.Schauer@mytum.de

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Pedro Alves]

On 08/24/2015 03:59 PM, Peter Schauer wrote:

>>> This looks like a memory problem on the build slave, it seems that
>>> memory flipped from 0x73 ('s') to 0x63 ('c').
>>
>> OOC, how did you figure this out?
> 
> From these lines:
> 
> build-gnulib/import/stdlib.h:41:20: fatal error: /usr/lib/gcc/mips-linux-gnu/4.9/include/ctddef.h: No such file or directory
>   #include <stddef.h>
>                      ^
> 
> gcc lists the correct source include (stddef.h), but then complains
> about not finding ctddef.h, so it looks like memory flipped somewhere
> between (or while) reading the include file name and then trying to
> access the include file.
> 

Aaaah.  Nice catch.

Brendan, is anyone looking at sorting this out?

Thanks,
Pedro Alves

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Brendan Kirby]

On 08/24/2015 08:03 AM, Pedro Alves wrote:
> On 08/24/2015 03:59 PM, Peter Schauer wrote:
>
>>>> This looks like a memory problem on the build slave, it seems that
>>>> memory flipped from 0x73 ('s') to 0x63 ('c').
>>> OOC, how did you figure this out?
>> From these lines:
>>
>> build-gnulib/import/stdlib.h:41:20: fatal error: /usr/lib/gcc/mips-linux-gnu/4.9/include/ctddef.h: No such file or directory
>>   #include <stddef.h>
>>                      ^
>>
>> gcc lists the correct source include (stddef.h), but then complains
>> about not finding ctddef.h, so it looks like memory flipped somewhere
>> between (or while) reading the include file name and then trying to
>> access the include file.
>>
> Aaaah.  Nice catch.
>
> Brendan, is anyone looking at sorting this out?
I'm not aware of anyone working on this.  I'm CC'ing a few people who
might be.  If it turns out to be a hardware problem, then I can move it
to another Edge Router Pro.   But, I don't have anything other than Edge
Router Pro boards to put it on.

Brendan

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Pedro Alves]

On 08/24/2015 06:51 PM, Brendan Kirby wrote:

>> Brendan, is anyone looking at sorting this out?
> I'm not aware of anyone working on this.  I'm CC'ing a few people who
> might be.  If it turns out to be a hardware problem, then I can move it
> to another Edge Router Pro.   But, I don't have anything other than Edge
> Router Pro boards to put it on.

Thanks.

FYI, this triggers quite often.  For the past couple weeks alone,
I got personally nagged with spurious build failures 5 times.  And
of course everyone else has been getting their share as well.

Thanks,
Pedro Alves

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Sergio Durigan Junior]

On Monday, August 24 2015, Pedro Alves wrote:

> FYI, this triggers quite often.  For the past couple weeks alone,
> I got personally nagged with spurious build failures 5 times.  And
> of course everyone else has been getting their share as well.

I'm looking for a way to disable e-mail notifications for this specific
buildslave.  This is not trivial, but I'll see what I can do.

Thanks,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Pedro Alves]

On 08/24/2015 07:10 PM, Sergio Durigan Junior wrote:
> On Monday, August 24 2015, Pedro Alves wrote:
> 
>> FYI, this triggers quite often.  For the past couple weeks alone,
>> I got personally nagged with spurious build failures 5 times.  And
>> of course everyone else has been getting their share as well.
> 
> I'm looking for a way to disable e-mail notifications for this specific
> buildslave.  This is not trivial, but I'll see what I can do.

I think we should also disable the test regressions
notifications too.  If building is unreliable, so must testing be.
Maybe it's simpler to disable the builder until this is sorted out?

Thanks,
Pedro Alves

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Sergio Durigan Junior]

On Monday, August 24 2015, Pedro Alves wrote:

> On 08/24/2015 07:10 PM, Sergio Durigan Junior wrote:
>> On Monday, August 24 2015, Pedro Alves wrote:
>> 
>>> FYI, this triggers quite often.  For the past couple weeks alone,
>>> I got personally nagged with spurious build failures 5 times.  And
>>> of course everyone else has been getting their share as well.
>> 
>> I'm looking for a way to disable e-mail notifications for this specific
>> buildslave.  This is not trivial, but I'll see what I can do.
>
> I think we should also disable the test regressions
> notifications too.  If building is unreliable, so must testing be.
> Maybe it's simpler to disable the builder until this is sorted out?

Yeah, I thought about that.  I'll disable it for now, then.

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Sergio Durigan Junior]

On Monday, August 24 2015, I wrote:

> On Monday, August 24 2015, Pedro Alves wrote:
>
>> On 08/24/2015 07:10 PM, Sergio Durigan Junior wrote:
>>> On Monday, August 24 2015, Pedro Alves wrote:
>>> 
>>>> FYI, this triggers quite often.  For the past couple weeks alone,
>>>> I got personally nagged with spurious build failures 5 times.  And
>>>> of course everyone else has been getting their share as well.
>>> 
>>> I'm looking for a way to disable e-mail notifications for this specific
>>> buildslave.  This is not trivial, but I'll see what I can do.
>>
>> I think we should also disable the test regressions
>> notifications too.  If building is unreliable, so must testing be.
>> Maybe it's simpler to disable the builder until this is sorted out?
>
> Yeah, I thought about that.  I'll disable it for now, then.

Done.  The buildslave is still connected to the BuildBot, but there are
no builders assigned to it, and therefore no e-mails will be generated
by it.

FWIW, I think I figure out a way to disable e-mail notifications
(entirely) from a builder, so if Brendan wants to test other machines to
see if the problem is solved, it can be done.

Cheers,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Brendan Kirby]

On 08/24/2015 11:24 AM, Sergio Durigan Junior wrote:
> On Monday, August 24 2015, I wrote:
>
>> On Monday, August 24 2015, Pedro Alves wrote:
>>
>>> On 08/24/2015 07:10 PM, Sergio Durigan Junior wrote:
>>>> On Monday, August 24 2015, Pedro Alves wrote:
>>>>
>>>>> FYI, this triggers quite often.  For the past couple weeks alone,
>>>>> I got personally nagged with spurious build failures 5 times.  And
>>>>> of course everyone else has been getting their share as well.
>>>> I'm looking for a way to disable e-mail notifications for this specific
>>>> buildslave.  This is not trivial, but I'll see what I can do.
>>> I think we should also disable the test regressions
>>> notifications too.  If building is unreliable, so must testing be.
>>> Maybe it's simpler to disable the builder until this is sorted out?
>> Yeah, I thought about that.  I'll disable it for now, then.
> Done.  The buildslave is still connected to the BuildBot, but there are
> no builders assigned to it, and therefore no e-mails will be generated
> by it.
>
> FWIW, I think I figure out a way to disable e-mail notifications
> (entirely) from a builder, so if Brendan wants to test other machines to
> see if the problem is solved, it can be done.
I spoke with Rich and he thinks there might be a patch to the kernel
that may help with the earlier issues.  When I get that patch installed,
I'll let you know and we can turn it on again with e-mail notification
off to see how it does.

Brendan

Re: [PATCH obv/pushed] MIPS ptrace build fixes

[Sergio Durigan Junior]

On Monday, August 24 2015, Brendan Kirby wrote:

> On 08/24/2015 08:03 AM, Pedro Alves wrote:
>> On 08/24/2015 03:59 PM, Peter Schauer wrote:
>>
>>>>> This looks like a memory problem on the build slave, it seems that
>>>>> memory flipped from 0x73 ('s') to 0x63 ('c').
>>>> OOC, how did you figure this out?
>>> From these lines:
>>>
>>> build-gnulib/import/stdlib.h:41:20: fatal error: /usr/lib/gcc/mips-linux-gnu/4.9/include/ctddef.h: No such file or directory
>>>   #include <stddef.h>
>>>                      ^
>>>
>>> gcc lists the correct source include (stddef.h), but then complains
>>> about not finding ctddef.h, so it looks like memory flipped somewhere
>>> between (or while) reading the include file name and then trying to
>>> access the include file.
>>>
>> Aaaah.  Nice catch.
>>
>> Brendan, is anyone looking at sorting this out?
> I'm not aware of anyone working on this.  I'm CC'ing a few people who
> might be.  If it turns out to be a hardware problem, then I can move it
> to another Edge Router Pro.   But, I don't have anything other than Edge
> Router Pro boards to put it on.

Hi Brendan,

Another thing that has been worrying me is the fact that the MIPS
buildslave is awfully slow; it's been a while now that its list of
pending builds contains 60+ commits.  All the other buildslaves manage
to catch up one way or another, and their list of pending builds rarely
has more than 10 commits.

I was wondering if you could assign more slaves to perform the builds.
I know that may not be feasible for you, but that would be the easiest
solution to this problem.

Thanks,

-- 
Sergio
GPG key ID: 237A 54B1 0287 28BF 00EF  31F4 D0EB 7628 65FC 5E36
Please send encrypted e-mail if possible
http://sergiodj.net/