Ability to patch all loaded object files #9
Conversation
Ref: pmem#7 While executing ASAN builds, some unexpected syscalls are observed. These can not be predicted, so this patch attempts to filter them out. Also, the return value of the function named hook in test/hook_test_preload.c is changed to 7 from 99. This also helps with ASAN builds: ASAN reads the buffers passed write syscalls, after they return. Due to this, an ASAN error was triggered every time it tried to read dummy_data buffer's contents, which has fewer than 99 bytes.
GBuella
force-pushed
the
patch_em_all
branch
2 times, most recently
from
4b3a677 to
280abb2
Compare
When the INTERCEPT_ALL_OBJS environment variable is set (to any value), libsyscall_intercept attempts to patch all objects it finds. This is not (yet) documented, as it is only meant to be used for a few testing scenarios. It also has caveats: the objects to be patched are filter based on their names, thus if one libsyscall_intercept itself is not in loaded as a library called libsyscall_intercept.so, it will try to intercept itself. This can happen when it statically linked into some other object file.
|
Reviewed 19 of 20 files at r1. Comments from Reviewable |
|
For those who feel like some more explanation would be appropriate, it is going on in this PR is coming soon: https://github.com/GBuella/syscall_intercept/blob/patch_all_objs_2/hacking.md That hacking.md is going to describe things in greater detail, but it still needs some days to be written, so I plan to add it in another PR. |
|
Reviewed 1 of 20 files at r1. Comments from Reviewable |
|
Review status: Comments from Reviewable |
all files reviewed at latest revision, all discussions resolved, all commit checks successful.
|
LGTM Reviewed 20 of 20 files at r1. Comments from Reviewable |
#7
This change is