rebase segmentation fault

rebase segmentation fault

[Denis Excoffier]

Hello,

I have a Cygwin installation under Windows 7 (32 bits).

After 'rebase-trigger full' and 'Setup', autorebase seems to finish okay, but the /etc/rebase.db.i386 remains absent after that.
Not a real problem after all (since rebase was functioning well until some recent date and all is already rebased), but some
third party builds (like e.g. perl) insist on the presence of the /etc/rebase.db.i386 database.

Indeed, rebase segfaults in rebase.c and nothing is rebased. This occurs (see rebase-4.4.2) immediately after "Skip trailing
entries as long as there is no hole." because at some point, all img_info_list[end].base are equal to 0, the while() is getting
always true, and the variable 'end' is decremented by 1 at every iteration, until it reaches -1 where the segfault occurs.

I must confess that my system contains all the available packages (except for debuginfo ones), and the rebase_all contains 9275 lines.
Later, i uninstalled a few packages (a few categories in fact), and my rebase_all contained 8337 lines, rebase worked well and
/etc/rebase.db.i386 was present.

Question: how is 'rebase' supposed to notify the user that the list of files is too large (i.e. that the overall amount of DLL sizes
is more than 0x070000000)? In source code, i found no specific message to indicate this.

Regards,

Denis Excoffier.


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: rebase segmentation fault

[Eliot Moss]

On 7/16/2017 1:16 PM, Denis Excoffier wrote:
> Hello,
>
> I have a Cygwin installation under Windows 7 (32 bits).
>
> After 'rebase-trigger full' and 'Setup', autorebase seems to finish okay, but the /etc/rebase.db.i386 remains absent after that.
> Not a real problem after all (since rebase was functioning well until some recent date and all is already rebased), but some
> third party builds (like e.g. perl) insist on the presence of the /etc/rebase.db.i386 database.
>
> Indeed, rebase segfaults in rebase.c and nothing is rebased. This occurs (see rebase-4.4.2) immediately after "Skip trailing
> entries as long as there is no hole." because at some point, all img_info_list[end].base are equal to 0, the while() is getting
> always true, and the variable 'end' is decremented by 1 at every iteration, until it reaches -1 where the segfault occurs.
>
> I must confess that my system contains all the available packages (except for debuginfo ones), and the rebase_all contains 9275 lines.
> Later, i uninstalled a few packages (a few categories in fact), and my rebase_all contained 8337 lines, rebase worked well and
> /etc/rebase.db.i386 was present.
>
> Question: how is 'rebase' supposed to notify the user that the list of files is too large (i.e. that the overall amount of DLL sizes
> is more than 0x070000000)? In source code, i found no specific message to indicate this.

I would find notification of such overflow useful as well.

And once I have over-committed by installing too many packages,
I have found it rather difficult to figure out what to remove
that will actually help, given the general snarl of dependencies
of one thing on another ...  Maybe there is room for another tool
that works over the database and helps determine what can be done.

Regards - Eliot Moss

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: rebase segmentation fault

[David Stacey]

On 16/07/17 18:16, Denis Excoffier wrote:
> I have a Cygwin installation under Windows 7 (32 bits).
>
> [snip]
>
> I must confess that my system contains all the available packages (except for debuginfo ones), and the rebase_all contains 9275 lines.

Cygwin has grown to the size where a complete install of 32-bit Cygwin 
isn't possible - you'll exhaust all the space available for rebasing. 
Most people don't need a full install [1]; if you do for some reason 
then you'll have to switch to 64-bit Cygwin.

Dave.

[1] - https://stackoverflow.com/a/21233990 footnote 2 ;-)


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: rebase segmentation fault

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1068 bytes --]

On Jul 16 19:16, Denis Excoffier wrote:
> Hello,
> 
> I have a Cygwin installation under Windows 7 (32 bits).
> 
> After 'rebase-trigger full' and 'Setup', autorebase seems to finish
> okay, but the /etc/rebase.db.i386 remains absent after that.  Not a
> real problem after all (since rebase was functioning well until some
> recent date and all is already rebased), but some third party builds
> (like e.g. perl) insist on the presence of the /etc/rebase.db.i386
> database.
> 
> Indeed, rebase segfaults in rebase.c and nothing is rebased. This
> occurs (see rebase-4.4.2) immediately after "Skip trailing entries as
> long as there is no hole." because at some point, all
> img_info_list[end].base are equal to 0, the while() is getting always
> true, and the variable 'end' is decremented by 1 at every iteration,
> until it reaches -1 where the segfault occurs.

Patches welcome!


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: rebase segmentation fault

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1244 bytes --]

On Jul 17 10:35, Corinna Vinschen wrote:
> On Jul 16 19:16, Denis Excoffier wrote:
> > Hello,
> > 
> > I have a Cygwin installation under Windows 7 (32 bits).
> > 
> > After 'rebase-trigger full' and 'Setup', autorebase seems to finish
> > okay, but the /etc/rebase.db.i386 remains absent after that.  Not a
> > real problem after all (since rebase was functioning well until some
> > recent date and all is already rebased), but some third party builds
> > (like e.g. perl) insist on the presence of the /etc/rebase.db.i386
> > database.
> > 
> > Indeed, rebase segfaults in rebase.c and nothing is rebased. This
> > occurs (see rebase-4.4.2) immediately after "Skip trailing entries as
> > long as there is no hole." because at some point, all
> > img_info_list[end].base are equal to 0, the while() is getting always
> > true, and the variable 'end' is decremented by 1 at every iteration,
> > until it reaches -1 where the segfault occurs.
> 
> Patches welcome!

Never mind:

https://sourceware.org/git/?p=cygwin-apps/rebase.git;a=commitdiff;h=a7d415a


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

Re: rebase segmentation fault

[Erik Bray]

On Mon, Jul 17, 2017 at 11:05 AM, Corinna Vinschen wrote:
> On Jul 17 10:35, Corinna Vinschen wrote:
>> On Jul 16 19:16, Denis Excoffier wrote:
>> > Hello,
>> >
>> > I have a Cygwin installation under Windows 7 (32 bits).
>> >
>> > After 'rebase-trigger full' and 'Setup', autorebase seems to finish
>> > okay, but the /etc/rebase.db.i386 remains absent after that.  Not a
>> > real problem after all (since rebase was functioning well until some
>> > recent date and all is already rebased), but some third party builds
>> > (like e.g. perl) insist on the presence of the /etc/rebase.db.i386
>> > database.
>> >
>> > Indeed, rebase segfaults in rebase.c and nothing is rebased. This
>> > occurs (see rebase-4.4.2) immediately after "Skip trailing entries as
>> > long as there is no hole." because at some point, all
>> > img_info_list[end].base are equal to 0, the while() is getting always
>> > true, and the variable 'end' is decremented by 1 at every iteration,
>> > until it reaches -1 where the segfault occurs.
>>
>> Patches welcome!
>
> Never mind:
>
> https://sourceware.org/git/?p=cygwin-apps/rebase.git;a=commitdiff;h=a7d415a

Thanks! I've been bitten by this in the past as well, but wasn't
sufficiently motivated to chase down the fix once I knew what the
problem was :)

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: rebase segmentation fault

[Denis Excoffier]

> On 2017-07-17 11:05, Corinna Vinschen wrote:
> 
> On Jul 17 10:35, Corinna Vinschen wrote:
>> On Jul 16 19:16, Denis Excoffier wrote:
>>> Hello,
>>> 
>>> I have a Cygwin installation under Windows 7 (32 bits).
>>> 
>>> After 'rebase-trigger full' and 'Setup', autorebase seems to finish
>>> okay, but the /etc/rebase.db.i386 remains absent after that.  Not a
>>> real problem after all (since rebase was functioning well until some
>>> recent date and all is already rebased), but some third party builds
>>> (like e.g. perl) insist on the presence of the /etc/rebase.db.i386
>>> database.
>>> 
>>> Indeed, rebase segfaults in rebase.c and nothing is rebased. This
>>> occurs (see rebase-4.4.2) immediately after "Skip trailing entries as
>>> long as there is no hole." because at some point, all
>>> img_info_list[end].base are equal to 0, the while() is getting always
>>> true, and the variable 'end' is decremented by 1 at every iteration,
>>> until it reaches -1 where the segfault occurs.
>> 
>> Patches welcome!
> 
> Never mind:
> 
> https://sourceware.org/git/?p=cygwin-apps/rebase.git;a=commitdiff;h=a7d415a
> 

Hello,

Please someone to build a new version of the rebase package with this modification included?

Thanks in advance,

Denis Excoffier.


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple