xconfig segfault

xconfig segfault

[Randy Dunlap]

make xconfig is segfaulting on me in 2.6.19-rc6 and later
when I do ^F (find/search).
Works fine in 2.6.19-rc5 and earlier.

The only message log I get is:

qconf[5839]: segfault at 0000000000000008 rip 00000000004289bc rsp 00007fffa08ccf10 error 4

I don't see any changes in scripts/kconfig/* in 2.6.19-rc6.
Any ideas/suggestions?

Thanks,
---
~Randy

Re: xconfig segfault

[Adrian Bunk]

On Sun, Nov 19, 2006 at 04:12:31PM -0800, Randy Dunlap wrote:
> make xconfig is segfaulting on me in 2.6.19-rc6 and later
> when I do ^F (find/search).
> Works fine in 2.6.19-rc5 and earlier.
> 
> The only message log I get is:
> 
> qconf[5839]: segfault at 0000000000000008 rip 00000000004289bc rsp 00007fffa08ccf10 error 4
> 
> I don't see any changes in scripts/kconfig/* in 2.6.19-rc6.
> Any ideas/suggestions?

Works fine for me in -rc6.

Did you upgrade Qt, or could there be any other local change that broke 
it for you?

> Thanks,
> ~Randy

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

Re: xconfig segfault

[Randy Dunlap]

Adrian Bunk wrote:
> On Sun, Nov 19, 2006 at 04:12:31PM -0800, Randy Dunlap wrote:
>> make xconfig is segfaulting on me in 2.6.19-rc6 and later
>> when I do ^F (find/search).
>> Works fine in 2.6.19-rc5 and earlier.
>>
>> The only message log I get is:
>>
>> qconf[5839]: segfault at 0000000000000008 rip 00000000004289bc rsp 00007fffa08ccf10 error 4
>>
>> I don't see any changes in scripts/kconfig/* in 2.6.19-rc6.
>> Any ideas/suggestions?
> 
> Works fine for me in -rc6.
> 
> Did you upgrade Qt, or could there be any other local change that broke 
> it for you?

Didn't upgrade Qt.  I started out suspecting that it was a local change
that broke it, like a library, but I rebuilt/re-tested 2.6.19-rc[123456]
and rc1..rc5 all work for me, while rc6 segfaults.
And rc6 works for me on an i386/i686 machine, but fails on x86_64.

-- 
~Randy

Re: xconfig segfault

[Randy Dunlap]

On Sun, 19 Nov 2006 16:47:03 -0800 Randy Dunlap wrote:

> Adrian Bunk wrote:
> > On Sun, Nov 19, 2006 at 04:12:31PM -0800, Randy Dunlap wrote:
> >> make xconfig is segfaulting on me in 2.6.19-rc6 and later
> >> when I do ^F (find/search).
> >> Works fine in 2.6.19-rc5 and earlier.
> >>
> >> The only message log I get is:
> >>
> >> qconf[5839]: segfault at 0000000000000008 rip 00000000004289bc rsp 00007fffa08ccf10 error 4
> >>
> >> I don't see any changes in scripts/kconfig/* in 2.6.19-rc6.
> >> Any ideas/suggestions?
> > 
> > Works fine for me in -rc6.
> > 
> > Did you upgrade Qt, or could there be any other local change that broke 
> > it for you?
> 
> Didn't upgrade Qt.  I started out suspecting that it was a local change
> that broke it, like a library, but I rebuilt/re-tested 2.6.19-rc[123456]
> and rc1..rc5 all work for me, while rc6 segfaults.
> And rc6 works for me on an i386/i686 machine, but fails on x86_64.

I found the problem patch, but not the root cause.

The xconfig segfault begins in 2.6.19-rc5-git3 (-git2 is OK).
A relatively simple Kconfig change causes it (but why?).

(Note:  The running kernel doesn't matter, just which kernel tree
is being viewed/config-ed.)

If I back out the patches below, -git3 (xconfig ^F find/search)
works for me.

---
~Randy


diff -Naurp -X linux-2.6.19-rc5-git2/Documentation/dontdiff linux-2.6.19-rc5-git2/drivers/char/Kconfig linux-2.6.19-rc5-git3/drivers/char/Kconfig
--- linux-2.6.19-rc5-git2/drivers/char/Kconfig	2006-11-20 08:47:26.000000000 -0800
+++ linux-2.6.19-rc5-git3/drivers/char/Kconfig	2006-11-20 08:48:34.000000000 -0800
@@ -409,14 +409,6 @@ config SGI_MBCS
          If you have an SGI Altix with an attached SABrick
          say Y or M here, otherwise say N.
 
-config MSPEC
-	tristate "Memory special operations driver"
-	depends on IA64
-	help
-	  If you have an ia64 and you want to enable memory special
-	  operations support (formerly known as fetchop), say Y here,
-	  otherwise say N.
-
 source "drivers/serial/Kconfig"
 
 config UNIX98_PTYS
diff -Naurp -X linux-2.6.19-rc5-git2/Documentation/dontdiff linux-2.6.19-rc5-git2/arch/ia64/Kconfig linux-2.6.19-rc5-git3/arch/ia64/Kconfig
--- linux-2.6.19-rc5-git2/arch/ia64/Kconfig	2006-11-20 08:47:41.000000000 -0800
+++ linux-2.6.19-rc5-git3/arch/ia64/Kconfig	2006-11-20 08:48:34.000000000 -0800
@@ -484,6 +484,15 @@ source "net/Kconfig"
 
 source "drivers/Kconfig"
 
+config MSPEC
+	tristate "Memory special operations driver"
+	depends on IA64
+	select IA64_UNCACHED_ALLOCATOR
+	help
+	  If you have an ia64 and you want to enable memory special
+	  operations support (formerly known as fetchop), say Y here,
+	  otherwise say N.
+
 source "fs/Kconfig"
 
 source "lib/Kconfig"

Re: xconfig segfault

[Roman Zippel]

Hi,

On Mon, 20 Nov 2006, Randy Dunlap wrote:

> I found the problem patch, but not the root cause.
> 
> The xconfig segfault begins in 2.6.19-rc5-git3 (-git2 is OK).
> A relatively simple Kconfig change causes it (but why?).
> 
> (Note:  The running kernel doesn't matter, just which kernel tree
> is being viewed/config-ed.)
> 
> If I back out the patches below, -git3 (xconfig ^F find/search)
> works for me.

I cannot reproduce this. Could you try to run qconf within gdb for a 
backtrace (adding -g to HOST_EXTRACFLAGS might get more useful output).

bye, Roman

Re: xconfig segfault

[Randy Dunlap]

Roman Zippel wrote:
> Hi,
> 
> On Mon, 20 Nov 2006, Randy Dunlap wrote:
> 
>> I found the problem patch, but not the root cause.
>>
>> The xconfig segfault begins in 2.6.19-rc5-git3 (-git2 is OK).
>> A relatively simple Kconfig change causes it (but why?).
>>
>> (Note:  The running kernel doesn't matter, just which kernel tree
>> is being viewed/config-ed.)
>>
>> If I back out the patches below, -git3 (xconfig ^F find/search)
>> works for me.
> 
> I cannot reproduce this. Could you try to run qconf within gdb for a 
> backtrace (adding -g to HOST_EXTRACFLAGS might get more useful output).

Sure, let me know what you want next.

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 46953334203536 (LWP 23635)]
0x00000000004289bc in ConfigMainWindow::searchConfig ()
(gdb) bt
#0  0x00000000004289bc in ConfigMainWindow::searchConfig ()
#1  0x0000000000428ae2 in ConfigMainWindow::qt_invoke ()
#2  0x00002ab42ab8b79c in QObject::activate_signal ()
   from /usr/lib64/libqt-mt.so.3
#3  0x00002ab42ab8c4b3 in QObject::activate_signal ()
   from /usr/lib64/libqt-mt.so.3
#4  0x00002ab42ae889b3 in QAction::qt_invoke () from /usr/lib64/libqt-mt.so.3
#5  0x00002ab42ab8b79c in QObject::activate_signal ()
   from /usr/lib64/libqt-mt.so.3
#6  0x00002ab42ae6d628 in QSignal::signal () from /usr/lib64/libqt-mt.so.3
#7  0x00002ab42aba42a5 in QSignal::activate () from /usr/lib64/libqt-mt.so.3
#8  0x00002ab42ab31a45 in QAccelPrivate::activate ()
   from /usr/lib64/libqt-mt.so.3
#9  0x00002ab42ab333e2 in QAccelManager::dispatchAccelEvent ()
   from /usr/lib64/libqt-mt.so.3
#10 0x00002ab42ab338fc in qt_dispatchAccelEvent () from /usr/lib64/libqt-mt.so.3
#11 0x00002ab42ab35d7a in QApplication::notify () from /usr/lib64/libqt-mt.so.3
#12 0x00002ab42ab31fce in QAccelManager::tryAccelEvent ()
   from /usr/lib64/libqt-mt.so.3
#13 0x00002ab42ab323ac in qt_tryAccelEvent () from /usr/lib64/libqt-mt.so.3
#14 0x00002ab42aadb39c in QETWidget::translateKeyEvent ()
   from /usr/lib64/libqt-mt.so.3
#15 0x00002ab42aadc180 in QApplication::x11ProcessEvent ()
   from /usr/lib64/libqt-mt.so.3
#16 0x00002ab42aaeb22f in QEventLoop::processEvents ()
   from /usr/lib64/libqt-mt.so.3
#17 0x00002ab42ab49691 in QEventLoop::enterLoop () from /usr/lib64/libqt-mt.so.3
#18 0x00002ab42ab4953a in QEventLoop::exec () from /usr/lib64/libqt-mt.so.3
#19 0x0000000000425e23 in main ()
(gdb) 

-- 
~Randy

Re: xconfig segfault

[Roman Zippel]

Hi,

On Mon, 20 Nov 2006, Randy Dunlap wrote:

> > I cannot reproduce this. Could you try to run qconf within gdb for a
> > backtrace (adding -g to HOST_EXTRACFLAGS might get more useful output).
> 
> Sure, let me know what you want next.

Hmm, an uninitialize field is IMO the only thing that could go wrong here.

bye, Roman

---
 scripts/kconfig/qconf.cc |    1 +
 1 file changed, 1 insertion(+)

Index: linux-2.6/scripts/kconfig/qconf.cc
===================================================================
--- linux-2.6.orig/scripts/kconfig/qconf.cc
+++ linux-2.6/scripts/kconfig/qconf.cc
@@ -1259,6 +1259,7 @@ void ConfigSearchWindow::search(void)
  * Construct the complete config widget
  */
 ConfigMainWindow::ConfigMainWindow(void)
+	: searchWindow(0)
 {
 	QMenuBar* menu;
 	bool ok;

Re: xconfig segfault

[Randy Dunlap]

On Tue, 21 Nov 2006 00:30:36 +0100 (CET) Roman Zippel wrote:

> Hi,
> 
> On Mon, 20 Nov 2006, Randy Dunlap wrote:
> 
> > > I cannot reproduce this. Could you try to run qconf within gdb for a
> > > backtrace (adding -g to HOST_EXTRACFLAGS might get more useful output).
> > 
> > Sure, let me know what you want next.
> 
> Hmm, an uninitialize field is IMO the only thing that could go wrong here.

Great, that fixes it.  Thanks!

> bye, Roman
> 
> ---
>  scripts/kconfig/qconf.cc |    1 +
>  1 file changed, 1 insertion(+)
> 
> Index: linux-2.6/scripts/kconfig/qconf.cc
> ===================================================================
> --- linux-2.6.orig/scripts/kconfig/qconf.cc
> +++ linux-2.6/scripts/kconfig/qconf.cc
> @@ -1259,6 +1259,7 @@ void ConfigSearchWindow::search(void)
>   * Construct the complete config widget
>   */
>  ConfigMainWindow::ConfigMainWindow(void)
> +	: searchWindow(0)
>  {
>  	QMenuBar* menu;
>  	bool ok;
> -


---
~Randy