linux-kernel - [PATCH 2/2] x86/microcode/AMD: Handle multiple glued containers properly

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <20221219210656.5140-2-bp@alien8.de>
Date:   Mon, 19 Dec 2022 22:06:56 +0100
From:   Borislav Petkov <bp@...en8.de>
To:     X86 ML <x86@...nel.org>
Cc:     LKML <linux-kernel@...r.kernel.org>
Subject: [PATCH 2/2] x86/microcode/AMD: Handle multiple glued containers properly

From: Borislav Petkov <bp@...e.de>

It can happen that - especially during testing - the microcode
blobs of all families are all glued together in the initrd. The
current code doesn't check whether the current container matched
a microcode patch and continues to the next one, which leads to
save_microcode_in_initrd_amd() to look at the next and thus wrong one:

  microcode: parse_container: ucode: 0xffff88807e9d9082
  microcode: verify_patch: buf: 0xffff88807e9d90ce, buf_size: 26428
  microcode: verify_patch: proc_id: 0x8082, patch_fam: 0x17, this family: 0x17
  microcode: verify_patch: buf: 0xffff88807e9d9d56, buf_size: 23220
  microcode: verify_patch: proc_id: 0x8012, patch_fam: 0x17, this family: 0x17
  microcode: parse_container: MATCH: eq_id: 0x8012, patch proc_rev_id: 0x8012

<-- matching patch found

  microcode: verify_patch: buf: 0xffff88807e9da9de, buf_size: 20012
  microcode: verify_patch: proc_id: 0x8310, patch_fam: 0x17, this family: 0x17
  microcode: verify_patch: buf: 0xffff88807e9db666, buf_size: 16804
  microcode: Invalid type field (0x414d44) in container file section header.
  microcode: Patch section fail

<-- checking chokes on the microcode magic value of the next container.

  microcode: parse_container: saving container 0xffff88807e9d9082
  microcode: save_microcode_in_initrd_amd: scanned containers, data: 0xffff88807e9d9082, size: 9700a

and now if there's a next (and last container) it'll use that in
save_microcode_in_initrd_amd() and not find a proper patch, ofc.

Fix that by moving the out: label up, before the desc->mc check which
jots down the pointer of the matching patch and is used to signal to the
caller that it has found a matching patch in the current container.

Signed-off-by: Borislav Petkov <bp@...e.de>
---
 arch/x86/kernel/cpu/microcode/amd.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/arch/x86/kernel/cpu/microcode/amd.c b/arch/x86/kernel/cpu/microcode/amd.c
index 339c9666c8bc..d144f918a896 100644
--- a/arch/x86/kernel/cpu/microcode/amd.c
+++ b/arch/x86/kernel/cpu/microcode/amd.c
@@ -330,8 +330,9 @@ static size_t parse_container(u8 *ucode, size_t size, struct cont_desc *desc)
 		ret = verify_patch(x86_family(desc->cpuid_1_eax), buf, size, &patch_size, true);
 		if (ret < 0) {
 			/*
-			 * Patch verification failed, skip to the next
-			 * container, if there's one:
+			 * Patch verification failed, skip to the next container, if
+			 * there is one. Before exit, check whether that container has
+			 * found a patch already. If so, use it.
 			 */
 			goto out;
 		} else if (ret > 0) {
@@ -350,6 +351,7 @@ static size_t parse_container(u8 *ucode, size_t size, struct cont_desc *desc)
 		size -= patch_size + SECTION_HDR_SIZE;
 	}

+out:
 	/*
 	 * If we have found a patch (desc->mc), it means we're looking at the
 	 * container which has a patch for this CPU so return 0 to mean, @ucode
@@ -364,7 +366,6 @@ static size_t parse_container(u8 *ucode, size_t size, struct cont_desc *desc)
 		return 0;
 	}

-out:
 	return orig_size - size;
 }

-- 
2.35.1