Red Hat Bugzilla – Bug 212516
unshield md5 checksums always fail on x86_64
Last modified: 2007-11-30 17:11:46 EST
Description of problem: The x86_64 build of unshield is completely broken and unusable. It always reports an MD5 checksum failure for every single file in every cabinet you attempt to extract or test. Version-Release number of selected component (if applicable): unshield-0.5-3.fc5 How reproducible: Attempt to extract or test any InstallShield cabinet using unshield Actual results: unshield will report an MD5 checksum failure for every single file in the cabinet, and fail to extract the files Expected results: MD5 checksum failures should only occur if the cabinet is actually damaged Additional info: Googling for "unshield md5", I came across this: http://ports.openbsd.nu/archivers/unshield Note the commit dated 2006-10-18. Apparently, unshield contains its own implementation of md5 which is not 64-bit-clean code. The OpenBSD package maintainer fixed the issue by making unshield use the system md5 (OpenBSD has md5 in their libc) instead of its own implementation. We don't have that luxury, so it looks like we'll have to actually fix unshield's md5 (or rip it out and replace it with the one from coreutils, but I don't think they're license compatible)
*** Bug 212707 has been marked as a duplicate of this bug. ***
Created attachment 141419 [details] PATCH: fixing md5 sum code on 64 bit Hi, Since I've gotten quite good in fixing 64 bit bugs I've taken a quick look at this. Fixing it turns out to be rather trivial, take a look at the attached patch. If you're short on time I can drop this and an updated spec in CVS and requests builds if you want me to.
thanks for helping out with this :) pushed for fe-{3,4,5,6} and devel :)
p.s. 1) Don't forget to also send this upstream, so that other distros can pick this up eventually. 2) Offtopic, have you received my scummvm-tools mail? Is it ok for me to update it to the latest and submit to FE, or do you want todo that?