linux-mips.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Marcin Nowakowski <marcin.nowakowski@imgtec.com>
To: Ralf Baechle <ralf@linux-mips.org>
Cc: James Hogan <james.hogan@imgtec.com>, <linux-mips@linux-mips.org>
Subject: [PATCH v2 2/2] MIPS: ptrace: disable watchpoints if hit in kernel mode
Date: Mon, 23 Jan 2017 10:18:33 +0100	[thread overview]
Message-ID: <1485163113-21780-2-git-send-email-marcin.nowakowski@imgtec.com> (raw)
In-Reply-To: <1485163113-21780-1-git-send-email-marcin.nowakowski@imgtec.com>

If a watchpoint is hit when in kernel mode it is possible for the system
to end up in an infinite loop processing the watchpoint. This can happen
if a user sets a watchpoint in the kernel addess space (which is
possible in certain EVA configurations) or if a user sets a watchpoint
in a user area accessed directly by the kernel (eg. a user buffer
accessed via a syscall).

To prevent the infinite loop ensure that the watchpoint was hit in
userspace, and clear the watchpoint registers otherwise.

As this change could mean that a watchpoint is not hit when it should be
(when returning to the interrupted traced task on exception exit), the
resume_userspace path needs to be extended to conditionally restore the
watchpoint configuration. If a task switch occurs when returning to
userspace, the watchpoints will be restored in a typical way in the
switch_to() handler.

Signed-off-by: Marcin Nowakowski <marcin.nowakowski@imgtec.com>
---
 arch/mips/kernel/entry.S | 9 ++++++++-
 arch/mips/kernel/traps.c | 2 +-
 2 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/arch/mips/kernel/entry.S b/arch/mips/kernel/entry.S
index 7791840..1982fc6 100644
--- a/arch/mips/kernel/entry.S
+++ b/arch/mips/kernel/entry.S
@@ -55,7 +55,14 @@ resume_userspace:
 	LONG_L	a2, TI_FLAGS($28)	# current->work
 	andi	t0, a2, _TIF_WORK_MASK	# (ignoring syscall_trace)
 	bnez	t0, work_pending
-	j	restore_all
+#ifdef CONFIG_HARDWARE_WATCHPOINTS
+	li	t0, _TIF_LOAD_WATCH
+	and	t0, a2, t0
+	beqz	t0, 1f
+	PTR_L	a0, TI_TASK($28)
+	jal	mips_install_watch_registers
+#endif
+1:	j	restore_all
 
 #ifdef CONFIG_PREEMPT
 resume_kernel:
diff --git a/arch/mips/kernel/traps.c b/arch/mips/kernel/traps.c
index 6c7f9d7..5371cb0 100644
--- a/arch/mips/kernel/traps.c
+++ b/arch/mips/kernel/traps.c
@@ -1525,7 +1525,7 @@ asmlinkage void do_watch(struct pt_regs *regs)
 	 * their values and send SIGTRAP.  Otherwise another thread
 	 * left the registers set, clear them and continue.
 	 */
-	if (test_tsk_thread_flag(current, TIF_LOAD_WATCH)) {
+	if (user_mode(regs) && test_tsk_thread_flag(current, TIF_LOAD_WATCH)) {
 		mips_read_watch_registers();
 		local_irq_enable();
 		force_sig_info(SIGTRAP, &info, current);
-- 
2.7.4

WARNING: multiple messages have this Message-ID (diff)
From: Marcin Nowakowski <marcin.nowakowski@imgtec.com>
To: Ralf Baechle <ralf@linux-mips.org>
Cc: James Hogan <james.hogan@imgtec.com>, linux-mips@linux-mips.org
Subject: [PATCH v2 2/2] MIPS: ptrace: disable watchpoints if hit in kernel mode
Date: Mon, 23 Jan 2017 10:18:33 +0100	[thread overview]
Message-ID: <1485163113-21780-2-git-send-email-marcin.nowakowski@imgtec.com> (raw)
Message-ID: <20170123091833.vgpvN3MeMoDC3g-FZCi7sIU185K503_dyx2VWDiBznw@z> (raw)
In-Reply-To: <1485163113-21780-1-git-send-email-marcin.nowakowski@imgtec.com>

If a watchpoint is hit when in kernel mode it is possible for the system
to end up in an infinite loop processing the watchpoint. This can happen
if a user sets a watchpoint in the kernel addess space (which is
possible in certain EVA configurations) or if a user sets a watchpoint
in a user area accessed directly by the kernel (eg. a user buffer
accessed via a syscall).

To prevent the infinite loop ensure that the watchpoint was hit in
userspace, and clear the watchpoint registers otherwise.

As this change could mean that a watchpoint is not hit when it should be
(when returning to the interrupted traced task on exception exit), the
resume_userspace path needs to be extended to conditionally restore the
watchpoint configuration. If a task switch occurs when returning to
userspace, the watchpoints will be restored in a typical way in the
switch_to() handler.

Signed-off-by: Marcin Nowakowski <marcin.nowakowski@imgtec.com>
---
 arch/mips/kernel/entry.S | 9 ++++++++-
 arch/mips/kernel/traps.c | 2 +-
 2 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/arch/mips/kernel/entry.S b/arch/mips/kernel/entry.S
index 7791840..1982fc6 100644
--- a/arch/mips/kernel/entry.S
+++ b/arch/mips/kernel/entry.S
@@ -55,7 +55,14 @@ resume_userspace:
 	LONG_L	a2, TI_FLAGS($28)	# current->work
 	andi	t0, a2, _TIF_WORK_MASK	# (ignoring syscall_trace)
 	bnez	t0, work_pending
-	j	restore_all
+#ifdef CONFIG_HARDWARE_WATCHPOINTS
+	li	t0, _TIF_LOAD_WATCH
+	and	t0, a2, t0
+	beqz	t0, 1f
+	PTR_L	a0, TI_TASK($28)
+	jal	mips_install_watch_registers
+#endif
+1:	j	restore_all
 
 #ifdef CONFIG_PREEMPT
 resume_kernel:
diff --git a/arch/mips/kernel/traps.c b/arch/mips/kernel/traps.c
index 6c7f9d7..5371cb0 100644
--- a/arch/mips/kernel/traps.c
+++ b/arch/mips/kernel/traps.c
@@ -1525,7 +1525,7 @@ asmlinkage void do_watch(struct pt_regs *regs)
 	 * their values and send SIGTRAP.  Otherwise another thread
 	 * left the registers set, clear them and continue.
 	 */
-	if (test_tsk_thread_flag(current, TIF_LOAD_WATCH)) {
+	if (user_mode(regs) && test_tsk_thread_flag(current, TIF_LOAD_WATCH)) {
 		mips_read_watch_registers();
 		local_irq_enable();
 		force_sig_info(SIGTRAP, &info, current);
-- 
2.7.4

  parent reply	other threads:[~2017-01-23  9:19 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-01-23  9:18 [PATCH v2 1/2] MIPS: ptrace: disallow setting watchpoints in kernel address space Marcin Nowakowski
2017-01-23  9:18 ` Marcin Nowakowski
2017-01-23  9:18 ` Marcin Nowakowski [this message]
2017-01-23  9:18   ` [PATCH v2 2/2] MIPS: ptrace: disable watchpoints if hit in kernel mode Marcin Nowakowski
2017-01-24 17:09 ` [PATCH v2 1/2] MIPS: ptrace: disallow setting watchpoints in kernel address space Maciej W. Rozycki
2017-01-24 17:09   ` Maciej W. Rozycki
2017-01-24 18:54   ` James Hogan
2017-01-24 18:54     ` James Hogan
2017-01-24 20:52     ` Maciej W. Rozycki
2017-01-24 20:52       ` Maciej W. Rozycki
2017-01-24 22:05       ` James Hogan
2017-01-24 22:05         ` James Hogan
2017-01-24 23:07         ` Maciej W. Rozycki
2017-01-24 23:07           ` Maciej W. Rozycki
2017-01-25 14:39           ` Maciej W. Rozycki
2017-01-25 14:39             ` Maciej W. Rozycki

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=1485163113-21780-2-git-send-email-marcin.nowakowski@imgtec.com \
    --to=marcin.nowakowski@imgtec.com \
    --cc=james.hogan@imgtec.com \
    --cc=linux-mips@linux-mips.org \
    --cc=ralf@linux-mips.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;
as well as URLs for NNTP newsgroup(s).