Re: [RFC PATCH 00/16] PTI support for x86-32

[Pavel Machek]

On Fri 2018-01-19 12:07:26, Joerg Roedel wrote:
> Hey Pavel,
> 
> On Fri, Jan 19, 2018 at 11:55:28AM +0100, Pavel Machek wrote:
> > Thanks for doing the work.
> > 
> > I tried applying it on top of -next, and that did not succeed. Let me
> > try Linus tree...
> 
> Thanks for your help with testing this patch-set, but I recommend to
> wait for the next version, as review already found a couple of bugs that
> might crash your system. For example there are NMI cases that might
> crash your machine because the NMI happens in kernel mode before the cr3
> switch. VM86 mode is also definitly broken.

Thanks for heads-up. I guess I can disable NMI avoid VM86.

CONFIG_X86_PTDUMP_CORE should be responsible for boot fail. Disabling
it is not at all easy, as CONFIG_EMBEDDED selects CONFIG_EXPERTS
selects CONFIG_DEBUG_KERNEL selects CONFIG_X86_PTDUMP_CORE. (Crazy, if
you ask me). You may want to test with that enabled. Patch below might
fix it. (Signed-off-by: me).

Tests so far: kernel boots in qemu. Whole system boots on thinkpad
T40p, vulnerabities/meltdown says mitigation: PTI.. so I guess it
works.

Tested-by: me. :-)

Best regards,
								Pavel


diff --git a/arch/x86/mm/dump_pagetables.c b/arch/x86/mm/dump_pagetables.c
index 2a4849e..896b53b 100644
--- a/arch/x86/mm/dump_pagetables.c
+++ b/arch/x86/mm/dump_pagetables.c
@@ -543,7 +543,11 @@ EXPORT_SYMBOL_GPL(ptdump_walk_pgd_level_debugfs);
 static void ptdump_walk_user_pgd_level_checkwx(void)
 {
 #ifdef CONFIG_PAGE_TABLE_ISOLATION
+#ifdef CONFIG_X86_64
 	pgd_t *pgd = (pgd_t *) &init_top_pgt;
+#else
+	pgd_t *pgd = swapper_pg_dir;
+#endif
 
 	if (!static_cpu_has(X86_FEATURE_PTI))
 		return;

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Attachment: signature.asc
Description: Digital signature