On Tue, Jul 28, 2020 at 01:06:17PM +0200, Ingo Molnar wrote: > > * Arvind Sankar <[email protected]> wrote: > > > The first 7 patches are cleanup and minor bugfixes on the x86 KASLR > > code. > > > > The last one is a bit of an RFC. The memory regions used for KASLR are > > stored as 64-bit even on a 32-bit kernel. However there are still a few > > local variables that are 32-bit, but should be ok as far as I can see > > because they are assigned values that have been already limited to > > 32-bit. It does make it a little harder to verify that the code is > > correct. Since KASLR cannot actually use 64-bit regions for the 32-bit > > kernel, the patch reduces regions to their below-4G segment when > > creating them, making the actual variables 32-bit. Alternatively, the > > few local variables could be upgraded to 64-bit. > > > > v1->v2: > > - Fix a bug in the bugfix 5/8: overlap.start can be smaller than > > region.start, so shouldn't subtract before comparing. > > > > Arvind Sankar (8): > > x86/kaslr: Make command line handling safer > > x86/kaslr: Remove bogus warning and unnecessary goto > > x86/kaslr: Fix process_efi_entries comment > > x86/kaslr: Initialize mem_limit to the real maximum address > > x86/kaslr: Simplify __process_mem_region > > x86/kaslr: Simplify process_gb_huge_pages > > x86/kaslr: Clean up slot handling > > x86/kaslr: Don't use 64-bit mem_vector for 32-bit kernel > > > > arch/x86/boot/compressed/acpi.c | 7 +- > > arch/x86/boot/compressed/kaslr.c | 228 ++++++++++++------------------- > > arch/x86/boot/compressed/misc.h | 19 ++- > > 3 files changed, 111 insertions(+), 143 deletions(-) > > I've applied patches 1-4 to x86/kaslr and will push them out if they > pass testing - see the review feedback for the others. > > Thanks, > > Ingo
Ok, thanks, I will send them split out and replace the last patch.
