diff options
author | Samuel Bowman <sam@sambowman.tech> | 2022-08-20 21:55:55 -0400 |
---|---|---|
committer | Andreas Kling <kling@serenityos.org> | 2022-08-23 11:48:50 +0200 |
commit | 91574ed6778f7c7fb7e574d9bcdd87804103d04f (patch) | |
tree | dc3f0f2094bfe51c830465685443c35557fdc8a7 /Kernel/Syscalls/profiling.cpp | |
parent | 0abe4d8b97e825af5ce5a120c695dd299814039a (diff) | |
download | serenity-91574ed6778f7c7fb7e574d9bcdd87804103d04f.zip |
Kernel: Fix boot profiling
Boot profiling was previously broken due to init_stage2() passing the
event mask to sys$profiling_enable() via kernel pointer, but a user
pointer is expected.
To fix this, I added Process::profiling_enable() as an alternative to
Process::sys$profiling_enable which takes a u64 rather than a
Userspace<u64 const*>. It's a bit of a hack, but it works.
Diffstat (limited to 'Kernel/Syscalls/profiling.cpp')
-rw-r--r-- | Kernel/Syscalls/profiling.cpp | 7 |
1 files changed, 7 insertions, 0 deletions
diff --git a/Kernel/Syscalls/profiling.cpp b/Kernel/Syscalls/profiling.cpp index efba20f338..869ccf7dd2 100644 --- a/Kernel/Syscalls/profiling.cpp +++ b/Kernel/Syscalls/profiling.cpp @@ -24,6 +24,13 @@ ErrorOr<FlatPtr> Process::sys$profiling_enable(pid_t pid, Userspace<u64 const*> TRY(require_no_promises()); auto const event_mask = TRY(copy_typed_from_user(userspace_event_mask)); + return profiling_enable(pid, event_mask); +} + +// NOTE: This second entrypoint exists to allow the kernel to invoke the syscall to enable boot profiling. +ErrorOr<FlatPtr> Process::profiling_enable(pid_t pid, u64 event_mask) +{ + VERIFY_PROCESS_BIG_LOCK_ACQUIRED(this); if (pid == -1) { auto credentials = this->credentials(); |