From 0e58af4e1d2166e9e33375a0f121e4867010d4f8 Mon Sep 17 00:00:00 2001 From: Andy Lutomirski Date: Thu, 4 Dec 2014 16:48:17 -0800 Subject: [PATCH] x86/tls: Disallow unusual TLS segments Users have no business installing custom code segments into the GDT, and segments that are not present but are otherwise valid are a historical source of interesting attacks. For completeness, block attempts to set the L bit. (Prior to this patch, the L bit would have been silently dropped.) This is an ABI break. I've checked glibc, musl, and Wine, and none of them look like they'll have any trouble. Note to stable maintainers: this is a hardening patch that fixes no known bugs. Given the possibility of ABI issues, this probably shouldn't be backported quickly. Signed-off-by: Andy Lutomirski Acked-by: H. Peter Anvin Cc: stable@vger.kernel.org # optional Cc: Konrad Rzeszutek Wilk Cc: Linus Torvalds Cc: security@kernel.org Cc: Willy Tarreau Signed-off-by: Ingo Molnar --- arch/x86/kernel/tls.c | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) diff --git a/arch/x86/kernel/tls.c b/arch/x86/kernel/tls.c index e7650bd71109..3e551eee87b9 100644 --- a/arch/x86/kernel/tls.c +++ b/arch/x86/kernel/tls.c @@ -39,6 +39,28 @@ static bool tls_desc_okay(const struct user_desc *info) if (!info->seg_32bit) return false; + /* Only allow data segments in the TLS array. */ + if (info->contents > 1) + return false; + + /* + * Non-present segments with DPL 3 present an interesting attack + * surface. The kernel should handle such segments correctly, + * but TLS is very difficult to protect in a sandbox, so prevent + * such segments from being created. + * + * If userspace needs to remove a TLS entry, it can still delete + * it outright. + */ + if (info->seg_not_present) + return false; + +#ifdef CONFIG_X86_64 + /* The L bit makes no sense for data. */ + if (info->lm) + return false; +#endif + return true; }