[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[PATCH] xen/livepatch: Fix .altinstructions safety checks


  • To: Xen-devel <xen-devel@xxxxxxxxxxxxxxxxxxxx>
  • From: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
  • Date: Sat, 15 Apr 2023 03:22:29 +0100
  • Authentication-results: esa1.hc3370-68.iphmx.com; dkim=none (message not signed) header.i=none
  • Cc: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>, Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>, Ross Lagerwall <ross.lagerwall@xxxxxxxxxx>
  • Delivery-date: Sat, 15 Apr 2023 02:23:13 +0000
  • Ironport-data: A9a23:XjWoIqBjFW9+hRVW/1/jw5YqxClBgxIJ4kV8jS/XYbTApD0q1mYHx zEfXmiBP/eNNDH0fox/Pd+w9UoH7Z7RztFgQQY4rX1jcSlH+JHPbTi7wuUcHAvJd5GeExg3h yk6QoOdRCzhZiaE/n9BCpC48T8nk/nOHuGmYAL9EngZbRd+Tys8gg5Ulec8g4p56fC0GArIs t7pyyHlEAbNNwVcbyRFuspvlDs15K6p4G9B4gRlDRx2lAS2e0c9Xcp3yZ6ZdxMUcqEMdsamS uDKyq2O/2+x13/B3fv8z94X2mVTKlLjFVDmZkh+AsBOsTAbzsAG6Y4pNeJ0VKtio27hc+ada jl6ncfYpQ8BZsUgkQmGOvVSO3kW0aZuoNcrLZUj2CA6IoKvn3bEmp1T4E8K0YIw0956B3kVr cckCQtdUD/AruOa5amec7w57igjBJGD0II3v3hhyXfSDOo8QICFSKLPjTNa9G5u3IYUR6+YP pdHL2M1N3wsYDUWUrsTILs4kP2lmT/UdDpApUjOjaE2/3LS3Ep6172F3N/9I4TXFJQExR/Az o7A11/yWTErG+aC83mAqiyBodfPxAirdp1HQdVU8dY12QbOlwT/EiY+RVa95PW0lEO6c9ZeM FAPvDojq7Ao806mRcW7WAe3yFaHuRgGUtYWDOw+6ymK0KPf5wvfDW8BJgOtc/R/6pVwH2Zzk AbUwZWwX2cHXKCppWy134+ZqyvpYRMpDFRSWwBHZAopxsLGmdRm5v7QdeqPAJJZn/WsR2Gpn 2vb8Xli71kApZVVjvvmpDgrlxrp/8GUFVBtu207S0r/tmtEiJiZi5tEALQxxdJJN86nQ1aIp xDocODOvblVXflheMFgKdjh/Y1FBN7falUweXY1Q/EcG82FohZPh7x47jBkP1tOOc0ZYzLva 0K7kVoPtMYPbSTxNPUtPt/Z5yEWIU/ITI2NaxwpRoAWPsgZmPGvp0mCmnJ8L0iyyRNxwMnTy L+QcNq2DGZyNJmLOAGeHr9HuZdyn3BW+I8mbcyjp/hR+ebENSH9pHZsGAfmU93VG4ve8ViEq 44Ca5rao/idOcWnChTqHUcoBQhiBRAG6Vre8KS7qsbrztJaJVwc
  • Ironport-hdrordr: A9a23:XXd1qqoY9/73zW/6R1Wxb+gaV5oneYIsimQD101hICG8cqSj+f xG+85rsSMc6QxhPk3I9urhBEDtex/hHP1OkOws1NWZLWrbUQKTRekIh+bfKlXbakrDH4VmtJ uIHZIQNDSJNykZsfrH
  • List-id: Xen developer discussion <xen-devel.lists.xenproject.org>

The prior check has && vs || mixups, making it tautologically false and thus
providing no safety at all.  There are boundary errors too.

First start with a comment describing how the .altinstructions and
.altinstr_replacement sections interact, and perform suitable cross-checking.

Second, rewrite the alt_instr loop entirely from scratch.  Origin sites have
non-zero size, and must be fully contained within .text.  Any non-zero sized
replacements must be fully contained within .altinstr_replacement.

Fixes: f8a10174e8b1 ("xsplice: Add support for alternatives")
Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
---
CC: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
CC: Ross Lagerwall <ross.lagerwall@xxxxxxxxxx>

As a further observation, .altinstr_replacement shouldn't survive beyond its
use in apply_alternatives(), but the disp32 relative references (for x86 at
least) in alt_instr force .altinstr_replacement to be close to the payload
while being applied.
---
 xen/common/livepatch.c | 66 ++++++++++++++++++++++++++++++++++++++----
 1 file changed, 60 insertions(+), 6 deletions(-)

diff --git a/xen/common/livepatch.c b/xen/common/livepatch.c
index 784fbd92e913..020a9648d5ba 100644
--- a/xen/common/livepatch.c
+++ b/xen/common/livepatch.c
@@ -803,28 +803,82 @@ static int prepare_payload(struct payload *payload,
     if ( sec )
     {
 #ifdef CONFIG_HAS_ALTERNATIVE
+        /*
+         * (As of April 2023), Alternatives are formed of:
+         * - An .altinstructions section with an array of struct alt_instr's.
+         * - An .altinstr_replacement section containing instructions bytes.
+         *
+         * An individual alt_instr contains:
+         * - An orig reference, pointing into .text with a nonzero length
+         * - A repl reference, pointing into .altinstr_replacement
+         *
+         * It is legal to have zero-length replacements, meaning it is legal
+         * for the .altinstr_replacement section to be empty too.  An
+         * implementation detail means that a zero-length replacement's repl
+         * reference will be the start of the .altinstr_replacement section.
+         */
+        const struct livepatch_elf_sec *repl_sec;
         struct alt_instr *a, *start, *end;
 
         if ( !section_ok(elf, sec, sizeof(*a)) )
             return -EINVAL;
 
+        /* Tolerate an empty .altinstructions section... */
+        if ( sec->sec->sh_size == 0 )
+            goto alt_done;
+
+        /* ... but otherwise, there needs to be something to alter... */
+        if ( payload->text_size == 0 )
+        {
+            printk(XENLOG_ERR LIVEPATCH "%s Alternatives provided, but no 
.text\n",
+                   elf->name);
+            return -EINVAL;
+        }
+
+        /* ... and something to be altered to. */
+        repl_sec = livepatch_elf_sec_by_name(elf, ".altinstr_replacement");
+        if ( !repl_sec )
+        {
+            printk(XENLOG_ERR LIVEPATCH "%s .altinstructions provided, but no 
.altinstr_replacement\n",
+                   elf->name);
+            return -EINVAL;
+        }
+
         start = sec->load_addr;
         end = sec->load_addr + sec->sec->sh_size;
 
         for ( a = start; a < end; a++ )
         {
-            const void *instr = ALT_ORIG_PTR(a);
-            const void *replacement = ALT_REPL_PTR(a);
+            const void *orig = ALT_ORIG_PTR(a);
+            const void *repl = ALT_REPL_PTR(a);
+
+            /* orig must be fully within .text. */
+            if ( orig               < payload->text_addr ||
+                 a->orig_len        > payload->text_size ||
+                 orig + a->orig_len > payload->text_addr + payload->text_size )
+            {
+                printk(XENLOG_ERR LIVEPATCH
+                       "%s Alternative orig %p+%#x outside payload text 
%p+%#lx\n",
+                       elf->name, orig, a->orig_len, payload->text_addr, 
payload->text_size);
+                return -EINVAL;
+            }
 
-            if ( (instr < region->start && instr >= region->end) ||
-                 (replacement < region->start && replacement >= region->end) )
+            /*
+             * repl must be fully within .altinstr_replacement, even if they
+             * happen to both have zero length.
+             */
+            if ( repl               < repl_sec->load_addr ||
+                 a->repl_len        > repl_sec->sec->sh_size ||
+                 repl + a->repl_len > repl_sec->load_addr + 
repl_sec->sec->sh_size )
             {
-                printk(XENLOG_ERR LIVEPATCH "%s Alt patching outside payload: 
%p\n",
-                       elf->name, instr);
+                printk(XENLOG_ERR LIVEPATCH
+                       "%s Alternative repl %p+%#x outside 
.altinstr_replacement %p+%#lx\n",
+                       elf->name, repl, a->repl_len, repl_sec->load_addr, 
repl_sec->sec->sh_size);
                 return -EINVAL;
             }
         }
         apply_alternatives(start, end);
+    alt_done:;
 #else
         printk(XENLOG_ERR LIVEPATCH "%s: We don't support alternative 
patching\n",
                elf->name);
-- 
2.30.2




 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.