Fixed runtime error and uninitialized variable usage

[wargio]

I have embedded libmspack in rizin and we had these issues raised by our CI

+../subprojects/libmspack/libmspack/mspack/cabd.c:1403:66: runtime error: left shift of 156 by 24 places cannot be represented in type 'int'
../subprojects/libmspack/libmspack/mspack/chmd.c: In function ‘chmd_extract’:
../subprojects/libmspack/libmspack/mspack/chmd.c:1139:10: error: ‘length’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
1139 |   length -= self->d->offset;
     |          ^~
../subprojects/libmspack/libmspack/mspack/chmd.c:1042:9: note: ‘length’ was declared here
1042 |   off_t length, offset;
     |         ^~~~~~
cc1: all warnings being treated as errors

If interested i have a working meson configuration for the project.

[kyz]

Thank you for incorporating libmspack into the rizin project, and thank you for the patch.

I made both changes directly to the source, rather than merge it.

• There are lots of places where UBSan complains and I fixed all of them in a6b8a22
• The warning is a false positive, I have a full walkthrough of the flow here: Compilier warnings with 1.9.1 #36 (comment)
• The warning is quelled in 0d13d86 (just before gcc's analysis fails, rather than turn off analysis completely by initialising in the declaration)

As far as build systems are concerned, libmspack will continue to use autotools as it allows supporting a wide range of esoteric systems in addition to modern Linux/BSDs, and there is also another large merge request in progress (#32) to introduce the CMake build system, mainly for better support on Windows. But thank you for the offer.

[wargio]

awesome. so it's just up to me to update to the latest commit. thanks for the effort.