Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
From: Christina Schimpe <christina.schimpe@intel.com>
To: gdb-patches@sourceware.org
Cc: thiago.bauermann@linaro.org, luis.machado@arm.com
Subject: [PATCH v5 07/12] gdb: amd64 linux coredump support with shadow stack.
Date: Sat, 28 Jun 2025 01:28:05 -0700	[thread overview]
Message-ID: <20250628082810.332526-8-christina.schimpe@intel.com> (raw)
In-Reply-To: <20250628082810.332526-1-christina.schimpe@intel.com>

Intel's Control-Flow Enforcement Technology (CET) provides the shadow
stack feature for the x86 architecture.

This commit adds support to write and read the shadow-stack node in
corefiles.  This helps debugging return address violations post-mortem.
The format is synced with the linux kernel commit "x86: Add PTRACE
interface for shadow stack".  As the linux kernel restricts shadow
stack support to 64-bit, apply the fix for amd64 only.

Co-Authored-By: Christina Schimpe <christina.schimpe@intel.com>

Reviewed-by: Thiago Jung Bauermann <thiago.bauermann@linaro.org>
---
 gdb/amd64-linux-tdep.c                        |  57 ++++++++-
 .../gdb.arch/amd64-shadow-stack-corefile.c    |  42 +++++++
 .../gdb.arch/amd64-shadow-stack-corefile.exp  | 110 ++++++++++++++++++
 3 files changed, 205 insertions(+), 4 deletions(-)
 create mode 100644 gdb/testsuite/gdb.arch/amd64-shadow-stack-corefile.c
 create mode 100644 gdb/testsuite/gdb.arch/amd64-shadow-stack-corefile.exp

diff --git a/gdb/amd64-linux-tdep.c b/gdb/amd64-linux-tdep.c
index edb7d8da6ab..9af7a41ea26 100644
--- a/gdb/amd64-linux-tdep.c
+++ b/gdb/amd64-linux-tdep.c
@@ -47,6 +47,7 @@
 #include "expop.h"
 #include "arch/amd64-linux-tdesc.h"
 #include "inferior.h"
+#include "x86-tdep.h"
 
 /* The syscall's XML filename for i386.  */
 #define XML_SYSCALL_FILENAME_AMD64 "syscalls/amd64-linux.xml"
@@ -1593,6 +1594,14 @@ amd64_linux_record_signal (struct gdbarch *gdbarch,
   return 0;
 }
 
+/* Get shadow stack pointer state from core dump.  */
+
+static bool
+amd64_linux_core_read_ssp_state_p (bfd *abfd)
+{
+  return bfd_get_section_by_name (abfd, ".reg-ssp") != nullptr;
+}
+
 /* Get Linux/x86 target description from core dump.  */
 
 static const struct target_desc *
@@ -1602,11 +1611,14 @@ amd64_linux_core_read_description (struct gdbarch *gdbarch,
 {
   /* Linux/x86-64.  */
   x86_xsave_layout layout;
-  uint64_t xcr0 = i386_linux_core_read_xsave_info (abfd, layout);
-  if (xcr0 == 0)
-    xcr0 = X86_XSTATE_SSE_MASK;
+  uint64_t xstate_bv_mask = i386_linux_core_read_xsave_info (abfd, layout);
+  if (xstate_bv_mask == 0)
+    xstate_bv_mask = X86_XSTATE_SSE_MASK;
+
+  if (amd64_linux_core_read_ssp_state_p (abfd))
+    xstate_bv_mask |= X86_XSTATE_CET_U;
 
-  return amd64_linux_read_description (xcr0 & X86_XSTATE_ALL_MASK,
+  return amd64_linux_read_description (xstate_bv_mask & X86_XSTATE_ALL_MASK,
 				       gdbarch_ptr_bit (gdbarch) == 32);
 }
 
@@ -1637,6 +1649,35 @@ static const struct regset amd64_linux_xstateregset =
     amd64_linux_collect_xstateregset
   };
 
+/* Supply shadow stack pointer register from SSP to the register cache
+   REGCACHE.  */
+
+static void
+amd64_linux_supply_ssp (const regset *regset,
+			regcache *regcache, int regnum,
+			const void *ssp, size_t len)
+{
+  x86_supply_ssp (regcache, *static_cast<const uint64_t *> (ssp));
+}
+
+/* Collect the shadow stack pointer register from the register cache
+   REGCACHE and store it in SSP.  */
+
+static void
+amd64_linux_collect_ssp (const regset *regset,
+			 const regcache *regcache, int regnum,
+			 void *ssp, size_t len)
+{
+  x86_collect_ssp (regcache, *static_cast<uint64_t *> (ssp));
+}
+
+/* Shadow stack pointer register.  */
+
+static const struct regset amd64_linux_ssp_register
+  {
+    NULL, amd64_linux_supply_ssp, amd64_linux_collect_ssp
+  };
+
 /* Iterate over core file register note sections.  */
 
 static void
@@ -1653,6 +1694,14 @@ amd64_linux_iterate_over_regset_sections (struct gdbarch *gdbarch,
     cb (".reg-xstate", tdep->xsave_layout.sizeof_xsave,
 	tdep->xsave_layout.sizeof_xsave, &amd64_linux_xstateregset,
 	"XSAVE extended state", cb_data);
+
+  /* SSP can be unavailable.  Thus, we need to check the register status
+     in case we write a core file (regcache != nullptr).  */
+  if (tdep->ssp_regnum > 0
+      && (regcache == nullptr
+	  || REG_VALID == regcache->get_register_status (tdep->ssp_regnum)))
+    cb (".reg-ssp", 8, 8, &amd64_linux_ssp_register,
+	"shadow stack pointer", cb_data);
 }
 
 /* The instruction sequences used in x86_64 machines for a
diff --git a/gdb/testsuite/gdb.arch/amd64-shadow-stack-corefile.c b/gdb/testsuite/gdb.arch/amd64-shadow-stack-corefile.c
new file mode 100644
index 00000000000..f078e33810d
--- /dev/null
+++ b/gdb/testsuite/gdb.arch/amd64-shadow-stack-corefile.c
@@ -0,0 +1,42 @@
+/* This test program is part of GDB, the GNU debugger.
+
+   Copyright 2025 Free Software Foundation, Inc.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
+
+#include <stdio.h>
+
+/* Call the return instruction before function epilogue to trigger a
+   control-flow exception.  */
+void
+function ()
+{
+  unsigned long ssp;
+  asm volatile("xor %0, %0; rdsspq %0" : "=r" (ssp));
+
+  /* Print ssp to stdout so that the testcase can capture it.  */
+  printf ("%p\n", (void *) ssp);
+  fflush (stdout);
+
+  /* Manually cause a control-flow exception by executing a return
+     instruction before function epilogue, so the address atop the stack
+     is not the return instruction.  */
+  __asm__ volatile ("ret\n");
+}
+
+int
+main (void)
+{
+  function (); /* Break here.  */
+}
diff --git a/gdb/testsuite/gdb.arch/amd64-shadow-stack-corefile.exp b/gdb/testsuite/gdb.arch/amd64-shadow-stack-corefile.exp
new file mode 100644
index 00000000000..8784fc3622c
--- /dev/null
+++ b/gdb/testsuite/gdb.arch/amd64-shadow-stack-corefile.exp
@@ -0,0 +1,110 @@
+# Copyright 2021-2024 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# Test the shadow stack pointer note in core dumps.
+
+require allow_ssp_tests
+
+standard_testfile
+
+proc check_core_file {core_filename saved_pl3_ssp} {
+    global decimal
+
+    # Load the core file.
+    if [gdb_test "core $core_filename" \
+	    [multi_line \
+		 "Core was generated by .*\\." \
+		 "Program terminated with signal SIGSEGV, Segmentation fault.*" \
+		 "#0  function \\(\\) at .*amd64-shadow-stack-corefile.c:$decimal" \
+		 "$decimal.*__asm__ volatile \\(\"ret\\\\n\"\\);"] \
+	    "load core file"] {
+	return
+    }
+
+    # Check the value of ssp in the core file.
+    gdb_test "print/x \$pl3_ssp" "\\$\[0-9\]+ = $saved_pl3_ssp" \
+	"pl3_ssp contents from core file $saved_pl3_ssp"
+}
+
+save_vars { ::env(GLIBC_TUNABLES) } {
+
+    append_environment GLIBC_TUNABLES "glibc.cpu.hwcaps" "SHSTK"
+
+    if { [prepare_for_testing "failed to prepare" $testfile $srcfile \
+	  {debug additional_flags="-fcf-protection=return"}] } {
+	return
+    }
+
+    set linespec ${srcfile}:[gdb_get_line_number "Break here"]
+
+    if ![runto $linespec] {
+	return
+    }
+
+    # Continue until a crash.  The line with the hex number is optional because
+    # it's printed by the test program, and doesn't appear in the Expect buffer
+    # when testing a remote target.
+    gdb_test "continue" \
+	[multi_line \
+	     "Continuing\\." \
+	     "($hex\r\n)?" \
+	     "Program received signal SIGSEGV, Segmentation fault.*" \
+	     "function \\(\\) at .*amd64-shadow-stack-corefile.c:$decimal" \
+	     {.*__asm__ volatile \("ret\\n"\);}] \
+    "continue to SIGSEGV"
+
+    set ssp_in_gcore [get_valueof "/x" "\$pl3_ssp" "*unknown*"]
+
+    # Generate the gcore core file.
+    set gcore_filename [standard_output_file "${testfile}.gcore"]
+    set gcore_generated [gdb_gcore_cmd "$gcore_filename" "generate gcore file"]
+
+    # Obtain an OS-generated core file.  Save test program output to
+    # ${binfile}.out.
+    set core_filename [core_find $binfile {} {} "${binfile}.out"]
+    set core_generated [expr {$core_filename != ""}]
+    set os_core_name "${binfile}.core"
+    remote_exec build "mv $core_filename $os_core_name"
+    set core_filename $os_core_name
+
+    # At this point we have a couple of core files, the gcore one generated by
+    # GDB and the one generated by the operating system.  Make sure GDB can
+    # read both correctly.
+
+    if {$gcore_generated} {
+	clean_restart $binfile
+
+	with_test_prefix "gcore corefile" {
+	    check_core_file $gcore_filename $ssp_in_gcore
+	}
+    } else {
+	fail "gcore corefile not generated"
+    }
+
+    if {$core_generated} {
+	clean_restart $binfile
+
+	with_test_prefix "OS corefile" {
+	    # Read ssp value from saved output of the test program.
+	    set out_id [open ${binfile}.out "r"]
+	    set ssp_in_gcore [gets $out_id]
+
+	    close $out_id
+	    check_core_file $core_filename $ssp_in_gcore
+	}
+    } else {
+	untested "OS corefile not generated"
+    }
+}
-- 
2.43.0


  parent reply	other threads:[~2025-06-28  8:33 UTC|newest]

Thread overview: 67+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-06-28  8:27 [PATCH v5 00/12] Add CET shadow stack support Christina Schimpe
2025-06-28  8:27 ` [PATCH v5 01/12] gdb, testsuite: Extend core_find procedure to save program output Christina Schimpe
2025-07-14 12:21   ` Andrew Burgess
2025-07-17 13:37     ` Schimpe, Christina
2025-06-28  8:28 ` [PATCH v5 02/12] gdbserver: Add optional runtime register set type Christina Schimpe
2025-06-28  8:28 ` [PATCH v5 03/12] gdbserver: Add assert in x86_linux_read_description Christina Schimpe
2025-06-28  8:28 ` [PATCH v5 04/12] gdb: Sync up x86-gcc-cpuid.h with cpuid.h from gcc 14 branch Christina Schimpe
2025-06-28  8:28 ` [PATCH v5 05/12] gdb, gdbserver: Use xstate_bv for target description creation on x86 Christina Schimpe
2025-07-14 13:52   ` Andrew Burgess
2025-07-15 10:28     ` Schimpe, Christina
2025-07-23 12:47       ` Schimpe, Christina
2025-08-05 13:47         ` Andrew Burgess
2025-06-28  8:28 ` [PATCH v5 06/12] gdb, gdbserver: Add support of Intel shadow stack pointer register Christina Schimpe
2025-07-25 12:49   ` Andrew Burgess
2025-07-25 15:03     ` Schimpe, Christina
2025-08-01 12:54       ` Schimpe, Christina
2025-08-05 13:57       ` Andrew Burgess
2025-08-06 19:53         ` Schimpe, Christina
2025-08-06 19:54           ` Schimpe, Christina
2025-08-07  3:17             ` Thiago Jung Bauermann
2025-08-14 11:39           ` Andrew Burgess
2025-07-29 13:51   ` Andrew Burgess
2025-08-01 12:40     ` Schimpe, Christina
2025-08-10 19:01   ` H.J. Lu
2025-08-10 20:07     ` Schimpe, Christina
2025-06-28  8:28 ` Christina Schimpe [this message]
2025-07-29 14:46   ` [PATCH v5 07/12] gdb: amd64 linux coredump support with shadow stack Andrew Burgess
2025-07-30  1:55     ` Thiago Jung Bauermann
2025-07-30 11:42       ` Schimpe, Christina
2025-08-04 15:28         ` Schimpe, Christina
2025-08-05  4:29           ` Thiago Jung Bauermann
2025-08-05 15:29             ` Schimpe, Christina
2025-08-06 20:52             ` Luis
2025-08-11 11:52               ` Schimpe, Christina
2025-08-04 12:45     ` Schimpe, Christina
2025-06-28  8:28 ` [PATCH v5 08/12] gdb: Handle shadow stack pointer register unwinding for amd64 linux Christina Schimpe
2025-07-30  9:58   ` Andrew Burgess
2025-07-30 12:06     ` Schimpe, Christina
2025-06-28  8:28 ` [PATCH v5 09/12] gdb, gdbarch: Enable inferior calls for shadow stack support Christina Schimpe
2025-07-30 10:42   ` Andrew Burgess
2025-06-28  8:28 ` [PATCH v5 10/12] gdb: Implement amd64 linux shadow stack support for inferior calls Christina Schimpe
2025-07-30 11:58   ` Andrew Burgess
2025-07-31 12:32     ` Schimpe, Christina
2025-06-28  8:28 ` [PATCH v5 11/12] gdb, gdbarch: Introduce gdbarch method to get the shadow stack pointer Christina Schimpe
2025-07-30 12:22   ` Andrew Burgess
2025-08-04 13:01     ` Schimpe, Christina
2025-08-14 15:50       ` Andrew Burgess
2025-08-19 15:37         ` Schimpe, Christina
2025-06-28  8:28 ` [PATCH v5 12/12] gdb: Enable displaced stepping with shadow stack on amd64 linux Christina Schimpe
2025-07-30 13:59   ` Andrew Burgess
2025-07-31 17:29     ` Schimpe, Christina
2025-07-08 15:18 ` [PATCH v5 00/12] Add CET shadow stack support Schimpe, Christina
2025-08-14  7:52   ` Schimpe, Christina
2025-07-11 10:36 ` Luis Machado
2025-07-11 13:54   ` Schimpe, Christina
2025-07-11 15:54     ` Luis Machado
2025-07-13 14:01       ` Schimpe, Christina
2025-07-13 19:05         ` Luis Machado
2025-07-13 19:57           ` Schimpe, Christina
2025-07-14  7:13           ` Luis Machado
2025-07-17 12:01             ` Schimpe, Christina
2025-07-17 14:59               ` Luis Machado
2025-07-23 12:45                 ` Schimpe, Christina
2025-07-28 17:05                   ` Luis Machado
2025-07-28 17:20                     ` Schimpe, Christina
2025-08-20  9:16 ` Schimpe, Christina
2025-08-20 15:21   ` Schimpe, Christina

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20250628082810.332526-8-christina.schimpe@intel.com \
    --to=christina.schimpe@intel.com \
    --cc=gdb-patches@sourceware.org \
    --cc=luis.machado@arm.com \
    --cc=thiago.bauermann@linaro.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox