Re: pc 0x25c9b0 in read in psymtab, but not in symtab

Re: pc 0x25c9b0 in read in psymtab, but not in symtab

[Roland Schwingel]

Hi...

gdb-owner@sourceware.org wrote on 11.01.2010 10:00:39:

 >
 > On Jan 10, 2010, at 3:41 AM, Chris Sutcliffe wrote:
 >
 > > Hi Tristan,
 > >
 > > I've been able to recreate the issue, but I've not been able to reduce
 > > it to a simple test case.  I can provide the dlls and exe that
 > > demonstrate the issue, if that would be helpful.
 >
 > Yes, it is.  Even if this is large, can you send them to me (privately) ?

I know this will not help you much, but I can reproduce it on windows 
(32bit) easily.
Unfortunately with a very big project where I cannot disclose the 
source. I already
tried to create an example where it happens but this will get complicated.

But maybe I can aid in debugging if you tell me what I can do I spend some
time on it. (At present I have just removed the errormessage in gdb to 
have it
(somewhat) usable at all).

Roland

pc 0x25c9b0 in read in psymtab, but not in symtab

[Elmenthaler, Jens]

Hi

using gdb 7.0 on RedHat Enterprise Linux 5.3, I see the following warning for some programs, but not all:

	warning: (Internal error: pc 0x25c9b0 in read in psymtab, but not in symtab.)

I wouldn't care so much about the warning, but once it appears, the program startup becomes damn slow. This seems to be related to dynamically loading additional shared libraries, which we do early in the program.

I looked a bit into the code that issues that message, but I'm not quite sure whether I ran into a bug or simply have to live with that fact.

Anything I can/should do here? 6.8 does not show this issue, but well, the pretty printers are a heavy argument.

Jens.

Re: pc 0x25c9b0 in read in psymtab, but not in symtab

[Chris Sutcliffe]

> using gdb 7.0 on RedHat Enterprise Linux 5.3, I see the following warning for some programs, but not all:
>
>        warning: (Internal error: pc 0x25c9b0 in read in psymtab, but not in symtab.)

I have had the same complaint from several users using GDB under
MinGW.  Is it possible to suppress the warning with a command line
flag?

Chris

-- 
Chris Sutcliffe
http://emergedesktop.org

Re: pc 0x25c9b0 in read in psymtab, but not in symtab

[Tristan Gingold]

On Jan 8, 2010, at 12:41 PM, Chris Sutcliffe wrote:

>> using gdb 7.0 on RedHat Enterprise Linux 5.3, I see the following warning for some programs, but not all:
>> 
>>        warning: (Internal error: pc 0x25c9b0 in read in psymtab, but not in symtab.)
> 
> I have had the same complaint from several users using GDB under
> MinGW.  Is it possible to suppress the warning with a command line
> flag?

I don't think it make sense.  It is really an internal error (ie a bug).
I was working on that very recently.  Do you have a full bug report ?  I could investigate this.

Tristan.

Re: pc 0x25c9b0 in read in psymtab, but not in symtab

[Chris Sutcliffe]

>>> using gdb 7.0 on RedHat Enterprise Linux 5.3, I see the following warning for some programs, but not all:
>>>
>>>        warning: (Internal error: pc 0x25c9b0 in read in psymtab, but not in symtab.)
>>
>> I have had the same complaint from several users using GDB under
>> MinGW.  Is it possible to suppress the warning with a command line
>> flag?
>
> I don't think it make sense.  It is really an internal error (ie a bug).
> I was working on that very recently.  Do you have a full bug report ?  I could investigate this.

I've had several users report the issue, but have not seen it myself.
Are you looking for a simple test case that exhibits this behaviour?
If so I can follow-up with these users and try to get one for you.

Chris

-- 
Chris Sutcliffe
http://emergedesktop.org

Re: pc 0x25c9b0 in read in psymtab, but not in symtab

[Tristan Gingold]

On Jan 8, 2010, at 12:51 PM, Chris Sutcliffe wrote:

>>>> using gdb 7.0 on RedHat Enterprise Linux 5.3, I see the following warning for some programs, but not all:
>>>> 
>>>>        warning: (Internal error: pc 0x25c9b0 in read in psymtab, but not in symtab.)
>>> 
>>> I have had the same complaint from several users using GDB under
>>> MinGW.  Is it possible to suppress the warning with a command line
>>> flag?
>> 
>> I don't think it make sense.  It is really an internal error (ie a bug).
>> I was working on that very recently.  Do you have a full bug report ?  I could investigate this.
> 
> I've had several users report the issue, but have not seen it myself.
> Are you looking for a simple test case that exhibits this behaviour?

Yes, that would be useful.

> If so I can follow-up with these users and try to get one for you.

Yes, please.

Tristan.

RE: pc 0x25c9b0 in read in psymtab, but not in symtab

[Elmenthaler, Jens]

> > I've had several users report the issue, but have not seen it myself.
> > Are you looking for a simple test case that exhibits this behaviour?

> Yes, that would be useful.

I will try to do the same. Though I'm not knowing yet how to start.

Jens.

-----Original Message-----
From: gdb-owner@sourceware.org [mailto:gdb-owner@sourceware.org] On Behalf Of Tristan Gingold
Sent: Freitag, 8. Januar 2010 12:57
To: Chris Sutcliffe
Cc: gdb@sourceware.org
Subject: Re: pc 0x25c9b0 in read in psymtab, but not in symtab


On Jan 8, 2010, at 12:51 PM, Chris Sutcliffe wrote:

>>>> using gdb 7.0 on RedHat Enterprise Linux 5.3, I see the following warning for some programs, but not all:
>>>> 
>>>>        warning: (Internal error: pc 0x25c9b0 in read in psymtab, but not in symtab.)
>>> 
>>> I have had the same complaint from several users using GDB under
>>> MinGW.  Is it possible to suppress the warning with a command line
>>> flag?
>> 
>> I don't think it make sense.  It is really an internal error (ie a bug).
>> I was working on that very recently.  Do you have a full bug report ?  I could investigate this.
> 
> I've had several users report the issue, but have not seen it myself.
> Are you looking for a simple test case that exhibits this behaviour?

Yes, that would be useful.

> If so I can follow-up with these users and try to get one for you.

Yes, please.

Tristan.

Re: pc 0x25c9b0 in read in psymtab, but not in symtab

[Chris Sutcliffe]

Hi Tristan,

>> I've had several users report the issue, but have not seen it myself.
>> Are you looking for a simple test case that exhibits this behaviour?
>
> Yes, that would be useful.

Apparently coming up with a simple test case isn't easy, but according
to Kai Tietz:

> yes this bug I encounter some time ago, too. It is related to DLL files
> not having any debugging information but are shown in backtrace. Here it
> warns once about psymtab != symtab and code in gdb fix it afterwards. IMHO
> this warning is simply pretty bogus here, or the DLL loader should
> generate for pe-coff the symtab, too.

Hopefully this is helpful.

If there is anything else I can do to help track down the bug, please
let me know.

Chris

-- 
Chris Sutcliffe
http://emergedesktop.org

Re: pc 0x25c9b0 in read in psymtab, but not in symtab

[Tristan Gingold]

On Jan 8, 2010, at 4:42 PM, Chris Sutcliffe wrote:

>> yes this bug I encounter some time ago, too. It is related to DLL files
>> not having any debugging information but are shown in backtrace. Here it
>> warns once about psymtab != symtab and code in gdb fix it afterwards. IMHO
>> this warning is simply pretty bogus here, or the DLL loader should
>> generate for pe-coff the symtab, too.
> 
> Hopefully this is helpful.

Not that much.  This doesn't sound coherent to me because if there is no debugging information both the
psymtab and the symtab should be empty.

> If there is anything else I can do to help track down the bug, please
> let me know.

It would be nice to have a reproducer.  It doesn't need to be small and I think sources are not needed
to investigate this issue.

Sadly, without reproducer I won't get far.

Tristan.

RE: pc 0x25c9b0 in read in psymtab, but not in symtab

[Elmenthaler, Jens]

> Sadly, without reproducer I won't get far.

In my case, the addresses complained about resolves to the following locations (RedHat Enterprise Linux 5.3):

maint translate-address 0x25c9b0
-> _dl_debug_state + 0 in section .text of /lib/ld-linux.so.2
maint translate-address 0x25c9b1
-> _dl_debug_state + 1 in section .text of /lib/ld-linux.so.2

This makes perfectly sense, as I see the messages while (or shortly after) the gdb is stopped because of shared library event.

I'm running out of time, right now. But a guess a program that dlopen's a shared library could be good candidate. I will give it a try as soon as possible.

Jens.

-----Original Message-----
From: gdb-owner@sourceware.org [mailto:gdb-owner@sourceware.org] On Behalf Of Tristan Gingold
Sent: Freitag, 8. Januar 2010 16:56
To: Chris Sutcliffe
Cc: gdb@sourceware.org
Subject: Re: pc 0x25c9b0 in read in psymtab, but not in symtab


On Jan 8, 2010, at 4:42 PM, Chris Sutcliffe wrote:

>> yes this bug I encounter some time ago, too. It is related to DLL files
>> not having any debugging information but are shown in backtrace. Here it
>> warns once about psymtab != symtab and code in gdb fix it afterwards. IMHO
>> this warning is simply pretty bogus here, or the DLL loader should
>> generate for pe-coff the symtab, too.
> 
> Hopefully this is helpful.

Not that much.  This doesn't sound coherent to me because if there is no debugging information both the
psymtab and the symtab should be empty.

> If there is anything else I can do to help track down the bug, please
> let me know.

It would be nice to have a reproducer.  It doesn't need to be small and I think sources are not needed
to investigate this issue.

Sadly, without reproducer I won't get far.

Tristan.

RE: pc 0x25c9b0 in read in psymtab, but not in symtab

[Elmenthaler, Jens]

> I'm running out of time, right now. But a guess a program that dlopen's a > shared library could be good candidate. I will give it a try as soon as < > possible.

I can no longer reproduce the problem: the reason can be one of the following thing:
- I rebooted my linux machine
- completely recompiled my software

I guess the trigger was the second, so I think the problem is sitting between keyboard and chair.

Jens.

-----Original Message-----
From: gdb-owner@sourceware.org [mailto:gdb-owner@sourceware.org] On Behalf Of Elmenthaler, Jens
Sent: Freitag, 8. Januar 2010 17:32
To: Tristan Gingold; Chris Sutcliffe
Cc: gdb@sourceware.org
Subject: RE: pc 0x25c9b0 in read in psymtab, but not in symtab

> Sadly, without reproducer I won't get far.

In my case, the addresses complained about resolves to the following locations (RedHat Enterprise Linux 5.3):

maint translate-address 0x25c9b0
-> _dl_debug_state + 0 in section .text of /lib/ld-linux.so.2
maint translate-address 0x25c9b1
-> _dl_debug_state + 1 in section .text of /lib/ld-linux.so.2

This makes perfectly sense, as I see the messages while (or shortly after) the gdb is stopped because of shared library event.

I'm running out of time, right now. But a guess a program that dlopen's a shared library could be good candidate. I will give it a try as soon as possible.

Jens.

-----Original Message-----
From: gdb-owner@sourceware.org [mailto:gdb-owner@sourceware.org] On Behalf Of Tristan Gingold
Sent: Freitag, 8. Januar 2010 16:56
To: Chris Sutcliffe
Cc: gdb@sourceware.org
Subject: Re: pc 0x25c9b0 in read in psymtab, but not in symtab


On Jan 8, 2010, at 4:42 PM, Chris Sutcliffe wrote:

>> yes this bug I encounter some time ago, too. It is related to DLL files
>> not having any debugging information but are shown in backtrace. Here it
>> warns once about psymtab != symtab and code in gdb fix it afterwards. IMHO
>> this warning is simply pretty bogus here, or the DLL loader should
>> generate for pe-coff the symtab, too.
> 
> Hopefully this is helpful.

Not that much.  This doesn't sound coherent to me because if there is no debugging information both the
psymtab and the symtab should be empty.

> If there is anything else I can do to help track down the bug, please
> let me know.

It would be nice to have a reproducer.  It doesn't need to be small and I think sources are not needed
to investigate this issue.

Sadly, without reproducer I won't get far.

Tristan.

Re: pc 0x25c9b0 in read in psymtab, but not in symtab

[Chris Sutcliffe]

Hi Tristan,

>> If there is anything else I can do to help track down the bug, please
>> let me know.
>
> It would be nice to have a reproducer.  It doesn't need to be small and I think sources are not needed
> to investigate this issue.
>
> Sadly, without reproducer I won't get far.

I'm working on trying to create a reproducer, but as I mentioned, I'm
not seeing the behaviour myself.  I have also asked for the users
reporting the issue to supply the code / exe / dll that exhibits the
behaviour.

I'll keep you posted.

Chris

-- 
Chris Sutcliffe
http://emergedesktop.org

Re: pc 0x25c9b0 in read in psymtab, but not in symtab

[Chris Sutcliffe]

Hi Tristan,

I've been able to recreate the issue, but I've not been able to reduce
it to a simple test case.  I can provide the dlls and exe that
demonstrate the issue, if that would be helpful.

Chris

-- 
Chris Sutcliffe
http://emergedesktop.org

Re: pc 0x25c9b0 in read in psymtab, but not in symtab

[Tristan Gingold]

On Jan 10, 2010, at 3:41 AM, Chris Sutcliffe wrote:

> Hi Tristan,
> 
> I've been able to recreate the issue, but I've not been able to reduce
> it to a simple test case.  I can provide the dlls and exe that
> demonstrate the issue, if that would be helpful.

Yes, it is.  Even if this is large, can you send them to me (privately) ?

Thanks,
Tristan.