Package: sslsplit Version: 0.5.0+dfsg-1 Severity: important Hi!
This is in reply to #848919 which you CCed to debian-mips. Sorry I've only just got around to looking at it. The underlying bug which seems to have caused #848919 is in the kh_dynbuf_hash_func in cachedsess.c: https://sources.debian.net/src/sslsplit/0.5.0%2Bdfsg-2/cachedsess.c/#L44 If rem != 0, h is only partially initialized by the memcpy and therefore the use of h on line 57 results in undefined behavior. Initializing h to 0 fixes the original FTBFS on mips. My guess is that your fix in 0.5.0+dfsg-2 which called fail_unless again setup some register which h happened to be allocated to so both invocations of cachemgr_dsess_set and cachemgr_dsess_get gave the same value by chance. I haven't looked into it much. It looks like this bug affects some of the other cache*.c files as well. Thanks, James
signature.asc
Description: OpenPGP digital signature