Skip to contentSkip to conversation

EFI urgent fix

GIT PULL New Matt Fleming 2 participants 3 messages Nov 4, 2015

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

Folks, the LKP robot reported an issue with Paolo's recent bug fix that syncs the identity mapping in 'initial_page_table'. Turns out that KERNEL_PGD_PTRS is not the correct constant to use when copying to the lower region because that's every PGD from PAGE_OFFSET to the end of the addressable memory.

Crucially, KERNEL_PGD_PTRS > KERNEL_PGD_BOUNDARY and so the patch ends up trashing some of the kernel mappings in 'initial_page'table, leading to boot crashes on 32-bit SMP when bringing APs online.

The following changes since commit 9ee870feaa9e0c6abef95a3b1fc518d88adfa2d3:

Merge branch 'x86/cpufeature' into x86/urgent, to pick up pending Intel MID change (2015-11-03 12:00:40 +0100)

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/mfleming/efi.git tags/efi-urgent

for you to fetch changes up to 5965d1bbeba70fe3626e4537f4729283cb0e75f7:

x86/setup: Fix recent boot crash on 32-bit SMP machines (2015-11-04 09:26:24 +0000)

----------------------------------------------------------------
* Avoid trashing the kernel mappings in 'initial_page_table' when copying the identity mapping from 'swapper_pg_dir'. This bug was introduced by a bug fix in v4.3 which erroneously copies too many entries from 'swapper_pg_dir'.

----------------------------------------------------------------
Matt Fleming (1):
x86/setup: Fix recent boot crash on 32-bit SMP machines

 arch/x86/kernel/setup.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
#1 lore raw

PATCH 1/1 x86/setup: Fix recent boot crash on 32-bit SMP machines

The LKP test robot reported that the bug fix in commit f5f3497cad8c ("x86/setup: Extend low identity map to cover whole kernel range") causes CONFIG_X86_32 SMP machines to crash on boot when trying to bring AP cpus online.

The above commit erroneously copies too many of the PGD entries to the low memory region of 'identity_page_table', resulting in some of the kernel mappings for PAGE_OFFSET being trashed because,

KERNEL_PGD_PTRS > KERNEL_PGD_BOUNDARY

The maximum number of PGD entries we can copy without corrupting the kernel mapping is KERNEL_PGD_BOUNDARY or pgd_index(PAGE_OFFSET).

Cc Paolo Bonzini <[email protected]>
Cc Laszlo Ersek <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Andy Lutomirski <[email protected]>
---
 arch/x86/kernel/setup.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/x86/kernel/setup.c +1 -1
index a3cccbfc5f77..2b8cbd693da8 100644
@@ -1180,7 +1180,7 @@ void __init setup_arch(char **cmdline_p)
 	 */
 	clone_pgd_range(initial_page_table,
 			swapper_pg_dir     + KERNEL_PGD_BOUNDARY,
-			KERNEL_PGD_PTRS);
+			KERNEL_PGD_BOUNDARY);
 #endif
 
 	tboot_probe();
Signature · 1 line
2.6.2
#2 lore raw
In reply to Matt Fleming
Quoted 3 lines · Matt Fleming

for you to fetch changes up to 5965d1bbeba70fe3626e4537f4729283cb0e75f7:

x86/setup: Fix recent boot crash on 32-bit SMP machines (2015-11-04 09:26:24 +0000)

I just picked that up manually :)