[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Requesting freeze exception for snappy 1.0.5-2



Hi,

I'm looking to see if it would be possible to get a freeze exception for
snappy (producing libsnappy1 and libsnappy-dev). testing has had 1.0.4-1 for
a long time; 1.0.5-1, a new upstream release (containing mostly performance
improvements) has been in unstable for about three months, but has failed to
compile on armel due to a bug specific to Debian ARM-systems (the only ARM
systems I use regularly are Ubuntu and Android systems), and no bug was filed
for this. Thus, it has not transitioned to testing, despite no bugs filed
against the package. I filed an FTBFS bug, #679964, myself when the DDPO
report arrived (just after the freeze!), for tracking purposes. 

The debdiff between 1.0.5-1 and 1.0.5-2 is simple, and I'll be putting it
upstream as soon as it goes through code review at work. Here it is:

--- snappy-1.0.5/debian/changelog
+++ snappy-1.0.5/debian/changelog
@@ -1,3 +1,11 @@
+snappy (1.0.5-2) unstable; urgency=low
+
+  * Disable unaligned accesses on ARMv4, not just ARMv5 and ARMv6.
+    Fixes unit test segfaults and subsequent FTBFS on armel.
+    (Closes: #679964) 
+
+ -- Steinar H. Gunderson <sesse@debian.org>  Mon, 02 Jul 2012 19:30:10 +0200
+
 snappy (1.0.5-1) unstable; urgency=low
 
   * New upstream release.
only in patch2:
unchanged:
--- snappy-1.0.5.orig/snappy-stubs-internal.h
+++ snappy-1.0.5/snappy-stubs-internal.h
@@ -200,6 +200,8 @@
 // This is a mess, but there's not much we can do about it.
 
 #elif defined(__arm__) && \
+      !defined(__ARM_ARCH_4__) && \
+      !defined(__ARM_ARCH_4T__) && \
       !defined(__ARM_ARCH_5__) && \
       !defined(__ARM_ARCH_5T__) && \
       !defined(__ARM_ARCH_5TE__) && \

1.0.5-2 has built successfully (which includes running an extensive unit
test battery, which was what caught the bug and stopped the package from
building on armel in the first place) on all release architectures,
including arm, armel and armhf.

Thanks for considering :-)

/* Steinar */
-- 
Homepage: http://www.sesse.net/


Reply to: