Re: [OS-BUILD PATCHv5 0/0] Add Partner Supported taint flag
From: Brian Foster on gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/merge_requests/1603#note_8889...
- [1/2] `f46fcbbf3fcf` ("Add Partner Supported taint flag")
A few questions/comments:
1. Any particular reason taint_flags[TAINT_PARTNER_SUPPORTED].module is
false? It's not immediately clear to me what output this flag affects, but it
seems to control whether the flag associates with the associated module one
way or another (see module_flags_taint()) and we do set mod->taints in the new
helper. Hm?
2. Maybe this is partly my dev environment, but I ran into a build failure
when compiling this locally:
fs/afs/main.c: In function ‘afs_init’:
fs/afs/main.c:205:9: error: implicit declaration of function
‘mark_partner_supported’ [-Werror=implicit-function-declaration]
205 | mark_partner_supported(KBUILD_MODNAME, THIS_MODULE);
However, I see that other internal helpers from this file do have header
declarations in kernel.h (see the associated CONFIG_RHEL_DIFFERENCES hunk).
Perhaps we should declare this new helper there as well?
Somewhat related, there's also an #else branch in kernel.h for the empty
compiled out variants. Using that should eliminate the need for the ifdef in
the second patch.
3. Kind of a nit and not sure this matters at all, but should we make any
assumptions about module type or license in the warning message? The code
comments say "Mark driver or kernel subsystem ...," but the warning explicitly
states "... GPL module ..." Maybe that text should be more generic..?