Login | Register
My pages Projects Community openCollabNet

Discussions > users > fsvs on Emdebian

fsvs
Discussion topic

Hide all messages in topic

All messages in topic

Re: fsvs on Emdebian

Author avathar
Full name Lukas Rötzer
Date 2010-05-06 08:31:51 PDT
Message Hi Phil!

Sorry for the long time that has passed, i was caught in the middle of
some exams on the university ;)

> Well, all the virtual filesystems (/dev/shm, /proc, /sys, usbfs etc.) should
> be excluded.
> Most of their entries don't behave like normal files, but are *shown* as
> files.
>
> Please take a look at the grouping/ignore patterns documentation
>
> http://doc.fsvs-soft​ware.org/doxygen-gif​/group__groups__spec​.html
>
> and look for "DEVICE:0".
>
>
thx to this i managed to import and checkout the filesystem with no
further problems..
many thanks for your help

regards
Lukas

Re: fsvs on Emdebian

Author pmarek
Full name P.Marek
Date 2010-04-27 22:37:40 PDT
Message Hello Lukas!

On Tuesday 27 April 2010 Lukas Rötzer wrote:
> I got emdebian testing running and compiled 1.2.2, but no luck..i got
> the same error as with 1.1.17 in the beginning (in direnum.c)
That's without --enable-debug, I assume?

But the differences between --enable-debug and without that are not that
many... strange.
 
> So i patched with your debugbatch and now i get a different error..here
> are the last lines of
>
> fsvs -d commit -m "initial":
...
> An error occurred at 20:44:39.560: Kein passender Prozess gefunden (3)
> in ci__nondir [commit.c:557]: svn_txdelta_send_stream: Can't read
> file 'proc/1/task/1/mem': Kein passender Prozess gefunden
> 20:44:39.676 main[fsvs.c:1422] memory stats: 0x76000 to 0x4db000, 4500 KB
>
> and the output from strace..
>
> open("./proc/1/task/1/mem", O_RDONLY|O_LARGEFILE|O_CLOEXEC) = 4
> fcntl64(4, F_GETFD) = 0x1 (flags FD_CLOEXEC)
> fcntl64(4, F_SETFD, FD_CLOEXEC) = 0
> poll([{fd=3, events=POLLIN}], 1, 0) = 0 (Timeout)
> read(4, 0x40f9c020, 102400) = -1 ESRCH (No such process)
Well, all the virtual filesystems (/dev/shm, /proc, /sys, usbfs etc.) should
be excluded.
Most of their entries don't behave like normal files, but are *shown* as
files.

Please take a look at the grouping/ignore patterns documentation

    http://doc.fsvs-soft​ware.org/doxygen-gif​/group__groups__spec​.html

and look for "DEVICE:0".


FSVS doesn't bring any default ignore patterns - they might be wrong, and they
would have to be different for every OS.


Regards,

Phil

--
Versioning your /etc, /home or even your whole installation?
             Try fsvs (fsvs.tigris.org)!

Re: fsvs on Emdebian

Author avathar
Full name Lukas Rötzer
Date 2010-04-27 06:52:40 PDT
Message Hi Phil!
>> debarm:/# ls -al /dev/initctl
>> prw------- 1 root root 0 1. Jan 1970 /dev/initctl
>>
> Regarding pipes a bug was fixed some time ago.
>
>
>
no..still on version 1.1.17..i'm still trying to get a newer version
running on emdebian..in the ppa you recently mentioned i only found
versions for i386 and x64, not for armel..i momentarily trying to get
emdebian testing (which has newer versions of subversion and so on)
running..

> PS: How about staying on the mailing list?
>
>
ah..sorry..is it enough to mail on users at fsvs dot tigris dot org to post in the
mailing list? cause i doesn't find our conversion there..

regards
Lukas

Re: fsvs on Emdebian

Author pmarek
Full name P.Marek
Date 2010-04-23 09:35:09 PDT
Message Hello Lukas,

>> Could you try the current version?
>
> I'll try to get the newest version running, but i think it's
> uncompatible to the svn version i use and some other things..
If you look at Gunnars packages you'll find some that work with subversion 1.4, too -
there's only a small patch needed (which is referenced somewhere in the ppa, too).

>> Either way here's a patch, that should tell me a bit
>> more what goes wrong.
>> Please apply, and send the "-d" output.
>>
>>
> the patchfile you sent me also works against 1.2.2,
> i modified it and
> applied it to version 1.1.17..the output is very
> long, so i send you
> only the last few lines..
...
> 16:55:06.823 ci___send_user_props​[commit.c:240] prop open: 2
> 16:55:06.823 ci__nondir[commit.c:413] initctl: status new, owner, group,
> mtime, umode; flags none
> 16:55:06.823 ci__nondir[commit.c:489] invalid/unknown file type 010600
> Speicherzugriffsfehler
>
> it seems though, that with the debug messages in place
> i get a totally
> different error message..
Yes, that's a completely different place.

Did you *only* change the sources, via the patch?
Not possibly another compiler?
You compiled the other version yourself, too?
Maybe some libraries changed?

What's the last output without -d?


Regards,

Phil

--
Versioning your /etc, /home or even your whole installation?
             Try fsvs (fsvs.tigris.org)!

Re: fsvs on Emdebian

Author pmarek
Full name P.Marek
Date 2010-04-22 22:19:06 PDT
Message Hello Lukas,

On Tuesday 20 April 2010 Gunnar Thielebein wrote:
> perhaps it is also possible for you trying to compile the package of recent
> version from ppa? https://launchpad.ne​t/~lorem-ipsum/+arch​ive/ppa
Could you try the current version?

Either way here's a patch, that should tell me a bit more what goes wrong.
Please apply, and send the "-d" output.

Thank you!


Regards,

Phil


--
Versioning your /etc, /home or even your whole installation?
             Try fsvs (fsvs.tigris.org)!
Attachments

Re: fsvs on Emdebian

Author tekknokra
Full name Gunnar Thielebein
Date 2010-04-20 02:17:22 PDT
Message Hi Lukas,

perhaps it is also possible for you trying to compile the package of recent
version from ppa? https://launchpad.ne​t/~lorem-ipsum/+arch​ive/ppa

IMO the build environment should be identical to Debian/Ubuntu environment so
changing the arch in control and distro in changelog and using

$ /usr/lib/pbuilder/pb​uilder-satisfydepend​s
$ debuild

should suffice? May you try that?

Best,
Gunnar

On 19.04.2010 11:59, Lukas Rötzer wrote:
> Am 2010-04-19 11:38, schrieb Philipp Marek:
>> Hello Lukas!
>>
>>
>>> I tried to get fsvs working on a NetusG20 ARM device with Emddebian installed.
>>>
>> ...
>>
>>> on execution i get the following error
>>>
>>> An error occurred at 17:25:00.848: Ungültige Adresse (14)
>>> in dir__enumerator: Speicherzugriffsfehler
>>>
>>> I searched quite a lot and also find a similar thread in these mailing archive,
>>> but there it seems, that the problem, which was related to the direnum.c, was
>>> resolved long time ago.
>>> On my Ubuntu machine it also works perfectly, so it seems it has something to do
>>> with the ARM Installation.
>>>
>> ...
>>
>>> Does anyone has an idea on this?
>>>
>> Could you please recompile (if necessary) with --enable-debug, and either
>> * run with gdb and send the backtrace, or
>> * start with -d, and send me the last few lines of output?
>>
>> Then I'll take a look.
>>
>> Maybe I'm doing some bad pointer arithmetic ... but it seems to work on all other
>> architectures, so probably not.
>>
>>
>> Regards,
>>
>> Phil
>>
>>
> Hi!
>
> Thanks for the fast answer.
>
> Recompiling will take some time, in the meanwhile a ran fsvs again with
> the -d parameter and the last few lines are the following..
>
> 10:46:44.919 dir__enumerator[direnum.c:465] found 85814 root
> 10:46:44.919 dir__enumerator[direnum.c:465] found 77921 bin
> 10:46:44.919 dir__enumerator[direnum.c:522] after loop found 21 entries,
> 110 bytes string-space
> 10:46:44.919 ops__allocate[est_ops.c:831] need 21 blocks, freelist=(nil)
> 10:46:44.920 ops__allocate[est_ops.c:869] no free list, allocating
> 10:46:44.920 ops__allocate[est_ops.c:886] giving 21 blocks at 0x69f88
> Speicherzugriffsfehler
>
> regards
>
> --------------------​--------------------​--------------
> http://fsvs.tigris.o​rg/ds/viewMessage.do​?dsForumId=3928&​dsMessageId=2589064
>
> To unsubscribe from this discussion, e-mail: [users-unsubscribe@f​svs.tigris.org].
>

Re: fsvs on Emdebian

Author pmarek
Full name P.Marek
Date 2010-04-19 23:46:58 PDT
Message Hello Lukas!

> Recompiling will take some time, in the meanwhile i ran fsvs again with
> the -d parameter and the last few lines are the following..
>
> 10:46:44.919 dir__enumerator[direnum.c:465] found 85814 root
> 10:46:44.919 dir__enumerator[direnum.c:465] found 77921 bin
> 10:46:44.919 dir__enumerator[direnum.c:522] after loop found 21 entries,
> 110 bytes string-space
> 10:46:44.919 ops__allocate[est_ops.c:831] need 21 blocks, freelist=(nil)
> 10:46:44.920 ops__allocate[est_ops.c:869] no free list, allocating
> 10:46:44.920 ops__allocate[est_ops.c:886] giving 21 blocks at 0x69f88
> Speicherzugriffsfehler
Hmmm, that doesn't help that much.

Do you have strace or gdb available?
If not, we'll have to put DEBUGP() lines in the dir__enumerator() until we find the
offending line.


Regards,

Phil

--
Versioning your /etc, /home or even your whole installation?
             Try fsvs (fsvs.tigris.org)!

Re: fsvs on Emdebian

Author avathar
Full name Lukas Rötzer
Date 2010-04-19 02:59:26 PDT
Message Am 2010-04-19 11:38, schrieb Philipp Marek:
> Hello Lukas!
>
>
>> I tried to get fsvs working on a NetusG20 ARM device with Emddebian installed.
>>
> ...
>
>> on execution i get the following error
>>
>> An error occurred at 17:25:00.848: Ungültige Adresse (14)
>> in dir__enumerator: Speicherzugriffsfehler
>>
>> I searched quite a lot and also find a similar thread in these mailing archive,
>> but there it seems, that the problem, which was related to the direnum.c, was
>> resolved long time ago.
>> On my Ubuntu machine it also works perfectly, so it seems it has something to do
>> with the ARM Installation.
>>
> ...
>
>> Does anyone has an idea on this?
>>
> Could you please recompile (if necessary) with --enable-debug, and either
> * run with gdb and send the backtrace, or
> * start with -d, and send me the last few lines of output?
>
> Then I'll take a look.
>
> Maybe I'm doing some bad pointer arithmetic ... but it seems to work on all other
> architectures, so probably not.
>
>
> Regards,
>
> Phil
>
>
Hi!

Thanks for the fast answer.

Recompiling will take some time, in the meanwhile a ran fsvs again with
the -d parameter and the last few lines are the following..

10:46:44.919 dir__enumerator[direnum.c:465] found 85814 root
10:46:44.919 dir__enumerator[direnum.c:465] found 77921 bin
10:46:44.919 dir__enumerator[direnum.c:522] after loop found 21 entries,
110 bytes string-space
10:46:44.919 ops__allocate[est_ops.c:831] need 21 blocks, freelist=(nil)
10:46:44.920 ops__allocate[est_ops.c:869] no free list, allocating
10:46:44.920 ops__allocate[est_ops.c:886] giving 21 blocks at 0x69f88
Speicherzugriffsfehler

regards

Re: fsvs on Emdebian

Author avathar
Full name Lukas Rötzer
Date 2010-04-19 02:49:24 PDT
Message Am 2010-04-19 11:38, schrieb Philipp Marek:
> Hello Lukas!
>
>
>> I tried to get fsvs working on a NetusG20 ARM device with Emddebian installed.
>>
> ...
>
>> on execution i get the following error
>>
>> An error occurred at 17:25:00.848: Ungültige Adresse (14)
>> in dir__enumerator: Speicherzugriffsfehler
>>
>> I searched quite a lot and also find a similar thread in these mailing archive,
>> but there it seems, that the problem, which was related to the direnum.c, was
>> resolved long time ago.
>> On my Ubuntu machine it also works perfectly, so it seems it has something to do
>> with the ARM Installation.
>>
> ...
>
>> Does anyone has an idea on this?
>>
> Could you please recompile (if necessary) with --enable-debug, and either
> * run with gdb and send the backtrace, or
> * start with -d, and send me the last few lines of output?
>
> Then I'll take a look.
>
> Maybe I'm doing some bad pointer arithmetic ... but it seems to work on all other
> architectures, so probably not.
>
>
> Regards,
>
> Phil
>
>
Hi!

Thanks for the fast answer.

Recompiling will take some time, in the meanwhile a ran fsvs again with
the -d parameter and the last few lines are the following..

10:46:44.919 dir__enumerator[direnum.c:465] found 85814 root
10:46:44.919 dir__enumerator[direnum.c:465] found 77921 bin
10:46:44.919 dir__enumerator[direnum.c:522] after loop found 21 entries,
110 bytes string-space
10:46:44.919 ops__allocate[est_ops.c:831] need 21 blocks, freelist=(nil)
10:46:44.920 ops__allocate[est_ops.c:869] no free list, allocating
10:46:44.920 ops__allocate[est_ops.c:886] giving 21 blocks at 0x69f88
Speicherzugriffsfehler

regards

Re: fsvs on Emdebian

Author pmarek
Full name P.Marek
Date 2010-04-19 02:38:43 PDT
Message Hello Lukas!

> I tried to get fsvs working on a NetusG20 ARM device with Emddebian installed.
...
> on execution i get the following error
>
> An error occurred at 17:25:00.848: Ungültige Adresse (14)
> in dir__enumerator: Speicherzugriffsfehler
>
> I searched quite a lot and also find a similar thread in these mailing archive,
> but there it seems, that the problem, which was related to the direnum.c, was
> resolved long time ago.
> On my Ubuntu machine it also works perfectly, so it seems it has something to do
> with the ARM Installation.
...
> Does anyone has an idea on this?
Could you please recompile (if necessary) with --enable-debug, and either
* run with gdb and send the backtrace, or
* start with -d, and send me the last few lines of output?

Then I'll take a look.

Maybe I'm doing some bad pointer arithmetic ... but it seems to work on all other
architectures, so probably not.


Regards,

Phil

--
Versioning your /etc, /home or even your whole installation?
             Try fsvs (fsvs.tigris.org)!

fsvs on Emdebian

Author avathar
Full name Lukas Rötzer
Date 2010-04-19 02:05:04 PDT
Message Hi!

I tried to get fsvs working on a NetusG20 ARM device with Emddebian installed.

The current version in the debian repositories is 1.1.16.
The Installation worked
fine, but on execution i get the following error

An error occurred at 17:25:00.848: Ungültige Adresse (14)
  in dir__enumerator: Speicherzugriffsfehler

I searched quite a lot and also find a similar thread in these mailing archive,
but there it seems, that the problem, which was related to the direnum.c, was
resolved long time ago.
On my Ubuntu machine it also works perfectly, so it seems it has something to do
with the ARM Installation.

I also compiled 1.1.17 and got the same error...also i was unable to compile
version 1.2.0 or greater due to various dependency failures, but i think they
would have the same problem anyway..

Does anyone has an idea on this?

kind regards

--
GMX.at - Österreichs FreeMail-Dienst mit über 2 Mio Mitgliedern
E-Mail, SMS & mehr! Kostenlos: http://portal.gmx.ne​t/de/go/atfreemail
Messages per page: