package/clamav: fix build with uClibc

clamav wants to use backtrace, and decides whether it can use it if it detects a glibc >= 2.1.

But uClibc does impersonate a glibc >= 2.1, so clamav concludes it is possible to use backtrace. So it includes execinfo.h, which is missing in our default uClibc config file.

So, just extend the test so that backtrace support is disable on uClibc, unless it has been configured with backtrace support.

A far better solution would be to add a ./configure check for backtrace, but this patch is sufficient enough.

Fixes:

    http://autobuild.buildroot.net/results/cff/cffa32fcedda735983d4805d6d4fa77844539b10/
    http://autobuild.buildroot.net/results/e0a/e0a765a94a538b0b936ea512f7aba0264fac6309/
    ...

---
 package/clamav/clamav-0003-backtrace-uClibc.patch | 22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)
 create mode 100644 package/clamav/clamav-0003-backtrace-uClibc.patch

new file mode 100644
index 0000000..82ad673
@@ -0,0 +1,22 @@
+mbox: do not use backtrace if using uClibc without backtrace support
+
+Since uClibc can be configured without support for backtrace, disable
+the backtrace if we are building with a uClibc that was built without
+backtrace.
+
+This is a bit hacky, and would greatly benefit from a test in ./configure
+instead, but does nicely as a quick fix for now.
+
+Signed-off-by: "Yann E. MORIN" <[email protected]>
+diff -durN clamav-0.98.4.orig/libclamav/mbox.c clamav-0.98.4/libclamav/mbox.c
+--- clamav-0.98.4.orig/libclamav/mbox.c	2014-05-21 17:25:05.000000000 +0200
++++ clamav-0.98.4/libclamav/mbox.c	2014-11-16 17:21:04.885383286 +0100
+@@ -83,7 +83,7 @@
+ #include <features.h>
+ #endif
+ 
+-#if __GLIBC__ == 2 && __GLIBC_MINOR__ >= 1
++#if __GLIBC__ == 2 && __GLIBC_MINOR__ >= 1 && !defined(__UCLIBC__) || defined(__UCLIBC_HAS_BACKTRACE__)
+ #define HAVE_BACKTRACE
+ #endif
+ #endif

1.9.1

"Yann E. MORIN" <[email protected]> wrote in news:1416155667-21771-1-git-send-email-yann.morin.1998 at free.fr:

Hi,

thanks, your patch supersedes http://patchwork.ozlabs.org/patch/406336/

Please add the upstream bugtracker URL to your patch:
https://bugzilla.clamav.net/show_bug.cgi?id=11170

Regards, Bernd

Bernd, All,

On 2014-11-16 18:21 +0100, Bernd Kuhls spake thusly:

Damn, it seems I should look in our patchwork before fixing bugs. ;-)

Thanks for pointing this out.

OK, I'll add it locally, but hopefully the maintainer can just stick this URL when he commits. :-p

Regards,
Yann E. MORIN.

.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

Dear Bernd Kuhls,

I believe Yann's patch is not an appropriate upstream solution, and a configure.ac check should be added instead.

Thomas

Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

Thomas, All,

On 2014-11-16 22:02 +0100, Thomas Petazzoni spake thusly:

That's what I think, too. But see what Peter had to say about that (as pointed out by Bernd):

    http://patchwork.ozlabs.org/patch/406336/

Now, upstream has both patches in their bug tracker: the original from Bernd that adds a check in ./configure (well, rather in m4/reorganization/headers.m4, but that gives a configure-time check), and my patch, does just slightly expands the current test to recognise uClibc.

It's now to upstream to decide what they want.

But back to Buildroot: we need a fix before the release, and that patch is what Peter asked for.

I don't care whichever we use, though. ;-)

Regards,
Yann E. MORIN.

.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

Just to clarify - I don't object against a configure script, my comment on http://patchwork.ozlabs.org/patch/406336/ was simply that the existing HAVE_BACKTRACE check didn't work for uClibc and it made more sense to fix that than add an additional HAVE_EXECINFO_H check.

Yeah, and I've committed it. When upstream decides (and perhaps makes a new release) we can update and get rid of the patch.

Bye, Peter Korsgaard

Well, you could certainly argue that a configure check would be nicer, but Yanns' fix is atleast in line with the upstream code (it just extends the existing glibc check in mbox.c).

Bye, Peter Korsgaard

Committed, thanks.

Bye, Peter Korsgaard