github-mirror/linux
1
0
Fork 0
mirror of https://github.com/torvalds/linux.git synced 2026-11-07 06:17:02 +01:00
Code Issues Actions Packages Projects Releases Wiki Activity Graph

um: mprotect() __init memory

Browse Source 
Unlike what the comment says, we could munmap() this (but
not reuse it for guest allocations), but then stray libc
allocations could technically conflict, so mprotect() it
to catch access bugs. Align it in the linker scripts too
so that all of it can be covered, not just some.

Link: https://patch.msgid.link/20260924084946.d118ec78e604.I993b671566a4e842dcc290b78f1c0004050362ba@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
This commit is contained in:
Johannes Berg 2026-09-24 08:49:45 +02:00
parent 8966ed0b2d
commit a418ab8fdd
3 changed files with 10 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
arch/um/kernel/dyn.lds.S
View File

@ -24,9 +24,8 @@ SECTIONS
INIT_TEXT_SECTION(PAGE_SIZE)
init.data : { INIT_DATA }
__init_end = .;
. = ALIGN(PAGE_SIZE);
__init_end = .;
/* Read-only sections, merged into text segment: */
.hash : { *(.hash) }

11
arch/um/kernel/mem.c
View File

@ -83,12 +83,17 @@ void __init arch_zone_limits_init(unsigned long *max_zone_pfns)
}
/*
* This can't do anything because nothing in the kernel image can be freed
* since it's not in kernel physical memory.
* We could munmap() this instead, but then libc allocations could
* land in this area and stray initdata access could erroneously
* succeed - just mprotect() it to reliably catch bad accesses.
*/
void free_initmem(void)
{
unsigned long start = PAGE_ALIGN((unsigned long)__init_begin);
unsigned long end = round_down((unsigned long)__init_end, PAGE_SIZE);
if (end > start)
os_protect_memory((void *)start, end - start, 0, 0, 0);
}
void *uml_kmalloc(int size, int flags)

1
arch/um/kernel/uml.lds.S
View File

@ -32,6 +32,7 @@ SECTIONS
INIT_TEXT_SECTION(0)
init.data : { INIT_DATA }
. = ALIGN(PAGE_SIZE);
__init_end = .;
.text :