MS office fails on wine due to fortify

The severity isn't critical; see http://bugs.winehq.org/page.cgi?id=fields.html#importance for correct usage. I don't see any evidence that the problem is related to ole32, so I'm changing the component field to unknown. Please perform a regression test:

http://wiki.winehq.org/RegressionTesting

I updated via Ubuntu USA Servers 3 days ago, and no new wine, I updated today and new wine version.

1.3.4

Whatever version was released to Ubuntu's update servers in the past 3 days, would be the version.

I'm not doing regression testing, sorry, not setup for it.

Marcus

(In reply to comment #2)
> I'm not doing regression testing, sorry, not setup for it.

If the instructions aren't clear, please visit http://forum.winehq.org to ask for help with the regression test.

Also affect me, same backtrace as OP.

ubuntu 10.10 wine ver 1.3.4

Created an attachment (id=31212)
backtrace-ves

backtraace for comparison

Install wine debug symbols (or build it yourself) and attach a +tid,+shell log.

Created an attachment (id=31213)
+tid,+shell log (compressed an otherwise too big log)

+tid,+shell log

(In reply to comment #7)
> Created an attachment (id=31213) [details]
> +tid,+shell log (compressed an otherwise too big log)
>
> +tid,+shell log

There's no crashes in this log.

I can't reproduce this, but I'm on openSUSE.

(In reply to comment #9)
> I can't reproduce this, but I'm on openSUSE.

Yes, looks like it's about ubuntu libc build:

---
/lib32/libc.so.6(__fortify_fail+0x50)[0xf7544a90]
/lib32/libc.so.6(+0xe39ca)[0xf75439ca]
/lib32/libc.so.6(__strcpy_chk+0x44)[0xf7542d44]
/usr/bin/../lib32/wine/shell32.dll.so(_ILCreateEntireNetwork+0x5c)[0x7e1e19ac]
/usr/bin/../lib32/wine/shell32.dll.so(+0x683c5)[0x7e20b3c5]
/usr/bin/../lib32/wine/shell32.dll.so(SHELL32_ParseNextElement+0x99)[0x7e21dcd9]
/usr/bin/../lib32/wine/shell32.dll.so(+0x60521)[0x7e203521]
---

There was a patch for this kind of the crash:
http://www.winehq.org/pipermail/wine-patches/2010-September/093377.html

(In reply to comment #11)
> There was a patch for this kind of the crash:
> http://www.winehq.org/pipermail/wine-patches/2010-September/093377.html

(In reply to comment #7)
> Created an attachment (id=31213) [details]
> +tid,+shell log (compressed an otherwise too big log)
>
> +tid,+shell log

It's a 'soft crash, meaning word catches the the exception, gives you the option to report about the crash and exits with the option in next run to recover the document (which works A-ok)

>
> It's a 'soft crash, meaning word catches the the exception, gives you the
> option to report about the crash and exits with the option in next run to
> recover the document (which works A-ok)

So thats great that its a soft crash, its still a crash to me, and I can't save any documents.

So since it happened after the upgrade and was fine before the upgrade, please tell me how I can fix it.

Thanks

Marcus

(In reply to comment #13)
> >
> > It's a 'soft crash, meaning word catches the the exception, gives you the
> > option to report about the crash and exits with the option in next run to
> > recover the document (which works A-ok)
>
> So thats great that its a soft crash, its still a crash to me, and I can't save
> any documents.
>
> So since it happened after the upgrade and was fine before the upgrade, please
> tell me how I can fix it.
>
> Thanks
>
> Marcus

If you're on ubuntu you can find the previous packages here :

10.10 (maveric)
https://launchpad.net/~ubuntu-wine/+archive/ppa/+packages?field.name_filter=&field.status_filter=superseded&field.series_filter=maverick

10.04 (lucid)
https://launchpad.net/~ubuntu-wine/+archive/ppa/+packages?field.name_filter=&field.status_filter=superseded&field.series_filter=lucid

previous version is 1.3.3 if I not mistaken

I can confirm this bug and that in fact the patch supplied by Mike Frysinger noted in comment 12 works perfectly.

x86_64 gentoo w/ wine 1.3.4.

(In reply to comment #15)
> I can confirm this bug and that in fact the patch supplied by Mike Frysinger
> noted in comment 12 works perfectly.
>
> x86_64 gentoo w/ wine 1.3.4.

Thanks Kyle.
How do I apply the patch?
Sorry to sound ignorant, but I'm still learning Ubuntu/Linux.

Marcus

Created an attachment (id=31235)
patch to fix the problem

This is just mikes patch with some white space fixups that don't come through on the email message in the link on comment 12.

(In reply to comment #17)
> Created an attachment (id=31235) [details]
> patch to fix the problem
>
> This is just mikes patch with some white space fixups that don't come through
> on the email message in the link on comment 12.

On Ubuntu this is going to be tricky, and almost impossible as there is no easy way to install your own custom version of wine (well there is, read up on it if you are interested). The best thing to do would be find out who packages wine for Ubuntu and point them to this bug, and specifically the patch here.

They will know what to do, and will rebuild the wine package and push it out through the normal channels (hopefully).

If you're feeling adventurous, download wine from git, and the patch above. Put the patch in the root of the wine/ git checkout and do a

patch -p1 < fix-msoffice.patch

when you are in the root directory of the wine/ git checkout at a terminal.

Best of luck (also, don't feel bad if you can't do this ... this is outside the scope of regular operating system users and should be taken care of by developers)

Why szName is defined with length of 1 if it gives an exception in certain situations in the first place?

(In reply to comment #19)
> Why szName is defined with length of 1 if it gives an exception in certain
> situations in the first place?

The code is perfectly correct, and doesn't give any exceptions unless it's built with broken crap like Fortify.

Yep, as Alexandre stated this bug disappears when -D_FORTIFY_SOURCE=0 is passed in CFLAGS. So yeah sorry for the noise and this bogus patch that fixes a non-existent problem.

On a side note, what the hell are distributions doing compiling wine with fortify source on? I'm on Gentoo right now and I trust their package maintainers more than other distributions and this is just worrisome.

Alexandre, would you say it is never a good idea to use -D_FORTIFY_SOURCE when compiling wine?

(In reply to comment #21)
> Alexandre, would you say it is never a good idea to use -D_FORTIFY_SOURCE when
> compiling wine?

It's definitely not a good idea. We already had to disable the broken longjmp checks that were causing crashes, and also to turn off some warnings because Fortify causes spurious ones. Now it looks like we need to disable the strcpy checks too. And with all the trouble it gave us, I have never seen it catch any real problem. I'm seriously considering turning it off globally in configure.

(In reply to comment #21)
> Yep, as Alexandre stated this bug disappears when -D_FORTIFY_SOURCE=0 is passed
> in CFLAGS. So yeah sorry for the noise and this bogus patch that fixes a
> non-existent problem.
>
> On a side note, what the hell are distributions doing compiling wine with
> fortify source on? I'm on Gentoo right now and I trust their package
> maintainers more than other distributions and this is just worrisome.
>

The dev packages that one of my computers use are maintained ( and compiled )by Scott Richie

It seems the problem has been acknowledged, the issue identified, and a reasonable, certain, and straightforward correction suggested. Why is this still marked Unconfirmed?

It's actually invalid, see the summary, and especially comment 20. I'm guessing we're holding it open to prevent a flood of duplicates.

I am hitting this bug on ubuntu maverick, the patch worked fine, but I am guessing that a lot of plp (without wine compile - patching knowledge) would be affected seriously by this bug.

As Jaime stated those (without wine compile - patching knowledge) would be seriously affected includes me.

I waited for the latest update to wine via the ubuntu channels, but nothing was fixed.

If someone could spell out the patch for the masses like me, who don't know how to patch, would be outstanding.

Thanks

(In reply to comment #27)
> If someone could spell out the patch for the masses like me, who don't know how
> to patch, would be outstanding.

It's not about patching Wine, it's about using certain compiler defines
which break the Wine binary. The Comment 21 contains the details. Just
file a bug with Ubuntu and ask them to compile Wine without Fortify.

On ubuntu i filled this bug:

https://bugs.launchpad.net/ubuntu/+source/wine1.2/+bug/665270

'Fixed' by http://source.winehq.org/git/wine.git/?a=commitdiff;h=105793bff75f6a869bd9ccfb390b2f88fa861235

Time to resolve this bug then.

Closing invalid.

Fixed in the last ubuntu release, Oct, 30th, 2010

*** Bug 25055 has been marked as a duplicate of this bug. ***

*** Bug 25073 has been marked as a duplicate of this bug. ***

*** Bug 25076 has been marked as a duplicate of this bug. ***

Nominating the configure change for 1.2.2

(In reply to comment #37)
> Nominating the configure change for 1.2.2

This bug was closed invalid, I'm not sure if it will be picked up for 1.2.2...

Removing 1.2.x milestone from bugs included in 1.2.2.