diff mbox series

[v6,2/2] xen/x86: Livepatch: support patching CET-enhanced functions

Message ID 81b800e254019ac3031e57307eea0ec422dc792f.1646897292.git.doebel@amazon.de (mailing list archive)
State New, archived
Headers show
Series None | expand

Commit Message

Doebel, Bjoern March 10, 2022, 7:35 a.m. UTC 
Xen enabled CET for supporting architectures. The control flow aspect of
CET expects functions that can be called indirectly (i.e., via function
pointers) to start with an ENDBR64 instruction. Otherwise a control flow
exception is raised.

This expectation breaks livepatching flows because we patch functions by
overwriting their first 5 bytes with a JMP + <offset>, thus breaking the
ENDBR64. We fix this by checking the start of a patched function for
being ENDBR64. In the positive case we move the livepatch JMP to start
behind the ENDBR64 instruction.

To avoid having to guess the ENDBR64 offset again on patch reversal
(which might race with other mechanisms adding/removing ENDBR
dynamically), use the livepatch metadata to store the computed offset
along with the saved bytes of the overwritten function.

Signed-off-by: Bjoern Doebel <doebel@amazon.de>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Ross Lagerwall <ross.lagerwall@citrix.com>
----
Changes since r1:
* (v2) use sizeof_field() to avoid unused variable warning
* (v3) make metadata variable const in arch_livepatch_revert
* (v4) rebase on top and make use of Andrew Cooper's was_endbr64() patch
* (v5) use padding byte to store offset rather than reducing opaque area
* (v6) style fixes
---
 xen/arch/x86/livepatch.c    | 39 +++++++++++++++++++++++++++++++------
 xen/include/public/sysctl.h |  3 ++-
 2 files changed, 35 insertions(+), 7 deletions(-)

Comments

Julien Grall March 11, 2022, 10:35 a.m. UTC | #1 
Hi Bjoern,

On 10/03/2022 07:35, Bjoern Doebel wrote:
> Xen enabled CET for supporting architectures. The control flow aspect of
> CET expects functions that can be called indirectly (i.e., via function
> pointers) to start with an ENDBR64 instruction. Otherwise a control flow
> exception is raised.
> 
> This expectation breaks livepatching flows because we patch functions by
> overwriting their first 5 bytes with a JMP + <offset>, thus breaking the
> ENDBR64. We fix this by checking the start of a patched function for
> being ENDBR64. In the positive case we move the livepatch JMP to start
> behind the ENDBR64 instruction.
> 
> To avoid having to guess the ENDBR64 offset again on patch reversal
> (which might race with other mechanisms adding/removing ENDBR
> dynamically), use the livepatch metadata to store the computed offset
> along with the saved bytes of the overwritten function.
> 
> Signed-off-by: Bjoern Doebel <doebel@amazon.de>
> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
> Reviewed-by: Ross Lagerwall <ross.lagerwall@citrix.com>
> ----
> Changes since r1:
> * (v2) use sizeof_field() to avoid unused variable warning
> * (v3) make metadata variable const in arch_livepatch_revert
> * (v4) rebase on top and make use of Andrew Cooper's was_endbr64() patch
> * (v5) use padding byte to store offset rather than reducing opaque area
> * (v6) style fixes

For the record, this patch depends on 
https://lore.kernel.org/xen-devel/20220308140126.8815-1-andrew.cooper3@citrix.com/.

Your patch will be merged once the prereq has been merged.

Cheers,
diff mbox series

Patch

diff --git a/xen/arch/x86/livepatch.c b/xen/arch/x86/livepatch.c
index 37c9b8435e..81c48fda7f 100644
--- a/xen/arch/x86/livepatch.c
+++ b/xen/arch/x86/livepatch.c
@@ -14,6 +14,7 @@ 
 #include <xen/vm_event.h>
 #include <xen/virtual_region.h>
 
+#include <asm/endbr.h>
 #include <asm/fixmap.h>
 #include <asm/nmi.h>
 #include <asm/livepatch.h>
@@ -114,8 +115,20 @@  int arch_livepatch_verify_func(const struct livepatch_func *func)
         if ( func->old_size < func->new_size )
             return -EINVAL;
     }
-    else if ( func->old_size < ARCH_PATCH_INSN_SIZE )
-        return -EINVAL;
+    else
+    {
+        /*
+         * Space needed now depends on whether the target function
+         * start{s,ed} with an ENDBR64 instruction.
+         */
+        uint8_t needed = ARCH_PATCH_INSN_SIZE;
+
+        if ( is_endbr64(func->old_addr) || was_endbr64(func->old_addr) )
+            needed += ENDBR64_LEN;
+
+        if ( func->old_size < needed )
+            return -EINVAL;
+    }
 
     return 0;
 }
@@ -130,12 +143,24 @@  void noinline arch_livepatch_apply(struct livepatch_func *func)
     uint8_t insn[sizeof(func->opaque)];
     unsigned int len;
 
+    func->patch_offset = 0;
     old_ptr = func->old_addr;
     len = livepatch_insn_len(func);
     if ( !len )
         return;
 
-    memcpy(func->opaque, old_ptr, len);
+    /*
+     * CET hotpatching support: We may have functions starting with an ENDBR64
+     * instruction that MUST remain the first instruction of the function,
+     * hence we need to move any hotpatch trampoline further into the function.
+     * For that we need to keep track of the patching offset used for any
+     * loaded hotpatch (to avoid racing against other fixups adding/removing
+     * ENDBR64 or similar instructions).
+     */
+    if ( is_endbr64(old_ptr) || was_endbr64(func->old_addr) )
+        func->patch_offset += ENDBR64_LEN;
+
+    memcpy(func->opaque, old_ptr + func->patch_offset, len);
     if ( func->new_addr )
     {
         int32_t val;
@@ -143,14 +168,15 @@  void noinline arch_livepatch_apply(struct livepatch_func *func)
         BUILD_BUG_ON(ARCH_PATCH_INSN_SIZE != (1 + sizeof(val)));
 
         insn[0] = 0xe9; /* Relative jump. */
-        val = func->new_addr - func->old_addr - ARCH_PATCH_INSN_SIZE;
+        val = func->new_addr - (func->old_addr + func->patch_offset +
+                                ARCH_PATCH_INSN_SIZE);
 
         memcpy(&insn[1], &val, sizeof(val));
     }
     else
         add_nops(insn, len);
 
-    memcpy(old_ptr, insn, len);
+    memcpy(old_ptr + func->patch_offset, insn, len);
 }
 
 /*
@@ -159,7 +185,8 @@  void noinline arch_livepatch_apply(struct livepatch_func *func)
  */
 void noinline arch_livepatch_revert(const struct livepatch_func *func)
 {
-    memcpy(func->old_addr, func->opaque, livepatch_insn_len(func));
+    memcpy(func->old_addr + func->patch_offset, func->opaque,
+           livepatch_insn_len(func));
 }
 
 /*
diff --git a/xen/include/public/sysctl.h b/xen/include/public/sysctl.h
index 55252e97f2..b0a4af8789 100644
--- a/xen/include/public/sysctl.h
+++ b/xen/include/public/sysctl.h
@@ -876,7 +876,8 @@  struct livepatch_func {
     uint8_t version;        /* MUST be LIVEPATCH_PAYLOAD_VERSION. */
     uint8_t opaque[LIVEPATCH_OPAQUE_SIZE];
     uint8_t applied;
-    uint8_t _pad[7];
+    uint8_t patch_offset;
+    uint8_t _pad[6];
     livepatch_expectation_t expect;
 };
 typedef struct livepatch_func livepatch_func_t;