RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-07-25 15:17:36 +00:00
Code Issues Activity Actions

LibCoreDump: Don't subtract one from the first stack frame's EIP

Browse source 
The first stack frame represents the current instruction pointer
rather than the return address so we shouldn't subtract one
from it.

Fixes #8162.
This commit is contained in:
Gunnar Beutner 2021-06-20 02:35:58 +02:00 • committed by Linus Groh
parent 995594b403
commit 25c73159ce
2 changed files with 12 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

7
Userland/Libraries/LibCoreDump/Backtrace.cpp
View file

@ -53,11 +53,16 @@ Backtrace::Backtrace(const Reader& coredump, const ELF::Core::ThreadInfo& thread
{
uint32_t* ebp = (uint32_t*)m_thread_info.regs.ebp;
uint32_t* eip = (uint32_t*)m_thread_info.regs.eip;
bool first_frame = true;
while (ebp && eip) {
// We use eip - 1 because the return address from a function frame
// is the instruction that comes after the 'call' instruction.
// However, because the first frame represents the faulting
// instruction rather than the return address we don't subtract
// 1 there.
VERIFY((FlatPtr)eip > 0);
add_entry(coredump, (FlatPtr)eip - 1);
add_entry(coredump, (FlatPtr)eip - (first_frame ? 0 : 1));
first_frame = false;
auto next_eip = coredump.peek_memory((FlatPtr)(ebp + 1));
auto next_ebp = coredump.peek_memory((FlatPtr)(ebp));
if (!next_eip.has_value() || !next_ebp.has_value())