LadybirdBrowser/ladybird
1
1
Fork 0
mirror of https://github.com/LadybirdBrowser/ladybird.git synced 2024-11-11 01:06:01 +03:00
Code Issues Packages Projects Releases Wiki Activity

Kernel: Always switch to own page tables when crashing/asserting

Browse Source 
I noticed this while debugging a crash in backtrace generation.
If a process would crash while temporarily inspecting another process's
address space, the crashing thread would still use the other process's
page tables while handling the crash, causing all kinds of confusion
when trying to walk the stack of the crashing thread.
This commit is contained in:
Andreas Kling 2020-01-19 09:54:58 +01:00
parent f7b394e9a1
commit 38fc31ff11
Notes: sideshowbarker 2024-07-19 09:57:51 +09:00 
Author: https://github.com/awesomekling
Commit: https://github.com/SerenityOS/serenity/commit/38fc31ff11f
1 changed files with 10 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
Kernel/Arch/i386/CPU.cpp
View File

@ -208,6 +208,10 @@ void handle_crash(RegisterDump& regs, const char* description, int signal)
hang();
}
// If a process crashed while inspecting another process,
// make sure we switch back to the right page tables.
MM.enter_process_paging_scope(current->process());
kprintf("\033[31;1mCRASH: %s. %s: %s(%u)\033[0m\n",
description,
current->process().is_ring0() ? "Kernel" : "Process",
@ -547,6 +551,12 @@ void __assertion_failed(const char* msg, const char* file, unsigned line, const
{
asm volatile("cli");
kprintf("ASSERTION FAILED: %s\n%s:%u in %s\n", msg, file, line, func);
// Switch back to the current process's page tables if there are any.
// Otherwise stack walking will be a disaster.
if (current)
MM.enter_process_paging_scope(current->process());
dump_backtrace();
asm volatile("hlt");
for (;;)