RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-25 23:47:45 +00:00
Code Issues Activity Actions

Kernel: Dispatch pending signals when returning from a syscall

Browse source 
It was quite easy to put the system into a heavy churn state by doing
e.g "cat /dev/zero".

It was then basically impossible to kill the "cat" process, even with
"kill -9", since signals are only delivered in two conditions:

a) The target thread is blocked in the kernel
b) The target thread is running in userspace

However, since "cat /dev/zero" command spends most of its time actively
running in the kernel, not blocked, the signal dispatch code just kept
postponing actually handling the signal indefinitely.

To fix this, we now check before returning from a syscall if there are
any pending unmasked signals, and if so, we take a dramatic pause by
blocking the current thread, knowing it will immediately be unblocked
by signal dispatch anyway. :^)
This commit is contained in:
Andreas Kling 2020-01-12 15:04:33 +01:00
parent 62a191b59a
commit 8b54ba0d61
2 changed files with 6 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

8
Kernel/Scheduler.cpp
View file

@ -322,11 +322,9 @@ bool Scheduler::pick_next()
if (&thread == current) if (&thread == current)
return IterationDecision::Continue; return IterationDecision::Continue;
// We know how to interrupt blocked processes, but if they are just executing // We know how to interrupt blocked processes, but if they are just executing
// at some random point in the kernel, let them continue. They'll be in userspace // at some random point in the kernel, let them continue.
// sooner or later and we can deliver the signal then. // Before returning to userspace from a syscall, we will block a thread if it has any
// FIXME: Maybe we could check when returning from a syscall if there's a pending // pending unmasked signals, allowing it to be dispatched then.
// signal and dispatch it then and there? Would that be doable without the
// syscall effectively being "interrupted" despite having completed?
if (thread.in_kernel() && !thread.is_blocked() && !thread.is_stopped()) if (thread.in_kernel() && !thread.is_blocked() && !thread.is_stopped())
return IterationDecision::Continue; return IterationDecision::Continue;
// NOTE: dispatch_one_pending_signal() may unblock the process. // NOTE: dispatch_one_pending_signal() may unblock the process.

3
Kernel/Syscall.cpp
View file

@ -135,4 +135,7 @@ void syscall_handler(RegisterDump regs)
// Check if we're supposed to return to userspace or just die. // Check if we're supposed to return to userspace or just die.
current->die_if_needed(); current->die_if_needed();
if (current->has_unmasked_pending_signals())
(void)current->block<Thread::SemiPermanentBlocker>(Thread::SemiPermanentBlocker::Reason::Signal);
} }