RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-05-31 04:58:13 +00:00
Code Issues Activity Actions

Kernel: Write-only regions should still be mapped as present

Browse source 
There is no real "read protection" on x86, so we have no choice but to
map write-only pages simply as "present & read/write".

If we get a read page fault in a non-readable region, that's still a
correctness issue, so we crash the process. It's by no means a complete
protection against invalid reads, since it's trivial to fool the kernel
by first causing a write fault in the same region.
This commit is contained in:
Andreas Kling 2020-01-20 13:06:55 +01:00
parent 4b7a89911c
commit 4ebff10bde
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
Kernel/VM/Region.cpp
View file

@ -259,7 +259,7 @@ void Region::map_individual_page_impl(size_t page_index)
} else {
pte.set_cache_disabled(!m_cacheable);
pte.set_physical_page_base(physical_page->paddr().get());
pte.set_present(is_readable());
pte.set_present(true);
if (should_cow(page_index))
pte.set_writable(false);
else
@ -331,7 +331,7 @@ PageFaultResponse Region::handle_fault(const PageFault& fault)
{
auto page_index_in_region = page_index_from_address(fault.vaddr());
if (fault.type() == PageFault::Type::PageNotPresent) {
if (!is_readable()) {
if (fault.is_read() && !is_readable()) {
dbgprintf("NP(non-readable) fault in Region{%p}[%u]\n", this, page_index_in_region);
return PageFaultResponse::ShouldCrash;
}