linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Oleg Nesterov <oleg@redhat.com>
To: Ananth Mavinakayanahalli <ananth@in.ibm.com>,
	Anton Arapov <arapov@gmail.com>,
	David Long <dave.long@linaro.org>,
	Denys Vlasenko <dvlasenk@redhat.com>,
	"Frank Ch. Eigler" <fche@redhat.com>,
	Ingo Molnar <mingo@kernel.org>, Jan Willeke <willeke@de.ibm.com>,
	Jim Keniston <jkenisto@us.ibm.com>,
	Mark Wielaard <mjw@redhat.com>,
	Pratyush Anand <panand@redhat.com>,
	Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Cc: linux-kernel@vger.kernel.org
Subject: [PATCH v2 08/11] uprobes: Change handle_trampoline() to flush the frames invalidated by longjmp()
Date: Tue, 7 Jul 2015 03:23:02 +0200	[thread overview]
Message-ID: <20150707012302.GA7540@redhat.com> (raw)
In-Reply-To: <20150707012210.GA7466@redhat.com>

Test-case:

	#include <stdio.h>
	#include <setjmp.h>

	jmp_buf jmp;

	void func_2(void)
	{
		longjmp(jmp, 1);
	}

	void func_1(void)
	{
		if (setjmp(jmp))
			return;
		func_2();
		printf("ERR!! I am running on the caller's stack\n");
	}

	int main(void)
	{
		func_1();
		return 0;
	}

fails if you probe func_1() and func_2() because handle_trampoline()
assumes that the probed function should must return and hit the bp
installed be prepare_uretprobe(). But in this case func_2() does not
return, so when func_1() returns the kernel uses the no longer valid
return_instance of func_2().

Change handle_trampoline() to unwind ->return_instances until we know
that the next chain is alive or NULL, this ensures that the current
chain is the last we need to report and free.

Alternatively, every return_instance could use unique trampoline_vaddr,
in this case we could use it as a key. And this could solve the problem
with sigaltstack() automatically.

But this approach needs more changes, and it puts the "hard" limit on
MAX_URETPROBE_DEPTH. Plus it can not solve another problem partially
fixed by the next patch.

Note: this change has no effect on !x86, the arch-agnostic version of
arch_uretprobe_is_alive() just returns "true".

TODO: as documented by the previous change, arch_uretprobe_is_alive()
can be fooled by sigaltstack/etc.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
---
 kernel/events/uprobes.c |   29 ++++++++++++++++++-----------
 1 files changed, 18 insertions(+), 11 deletions(-)

diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
index c5f316e..93d939c 100644
--- a/kernel/events/uprobes.c
+++ b/kernel/events/uprobes.c
@@ -1774,6 +1774,7 @@ static void handle_trampoline(struct pt_regs *regs)
 {
 	struct uprobe_task *utask;
 	struct return_instance *ri, *next;
+	bool valid;
 
 	utask = current->utask;
 	if (!utask)
@@ -1783,18 +1784,24 @@ static void handle_trampoline(struct pt_regs *regs)
 	if (!ri)
 		goto sigill;
 
-	next = find_next_ret_chain(ri);
-	/*
-	 * TODO: we should throw out return_instance's invalidated by
-	 * longjmp(), currently we assume that the probed function always
-	 * returns.
-	 */
-	instruction_pointer_set(regs, ri->orig_ret_vaddr);
 	do {
-		handle_uretprobe_chain(ri, regs);
-		ri = free_ret_instance(ri);
-		utask->depth--;
-	} while (ri != next);
+		/*
+		 * We should throw out the frames invalidated by longjmp().
+		 * If this chain is valid, then the next one should be alive
+		 * or NULL; the latter case means that nobody but ri->func
+		 * could hit this trampoline on return. TODO: sigaltstack().
+		 */
+		next = find_next_ret_chain(ri);
+		valid = !next || arch_uretprobe_is_alive(next, regs);
+
+		instruction_pointer_set(regs, ri->orig_ret_vaddr);
+		do {
+			if (valid)
+				handle_uretprobe_chain(ri, regs);
+			ri = free_ret_instance(ri);
+			utask->depth--;
+		} while (ri != next);
+	} while (!valid);
 
 	utask->return_instances = ri;
 	return;
-- 
1.5.5.1


  parent reply	other threads:[~2015-07-07  1:25 UTC|newest]

Thread overview: 30+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-07-07  1:22 [PATCH v2 00/11] uprobes: longjmp fixes Oleg Nesterov
2015-07-07  1:22 ` [PATCH v2 01/11] uprobes: Introduce get_uprobe() Oleg Nesterov
2015-07-07 12:44   ` Anton Arapov
2015-07-07  1:22 ` [PATCH v2 02/11] uprobes: Introduce free_ret_instance() Oleg Nesterov
2015-07-07 12:46   ` Anton Arapov
2015-07-07  1:22 ` [PATCH v2 03/11] uprobes: Send SIGILL if handle_trampoline() fails Oleg Nesterov
2015-07-07 12:51   ` Anton Arapov
2015-07-07  1:22 ` [PATCH v2 04/11] uprobes: Change prepare_uretprobe() to use uprobe_warn() Oleg Nesterov
2015-07-07 12:52   ` Anton Arapov
2015-07-07  1:22 ` [PATCH v2 05/11] uprobes: Change handle_trampoline() to find the next chain beforehand Oleg Nesterov
2015-07-07 12:54   ` Anton Arapov
2015-07-07  1:22 ` [PATCH v2 06/11] uprobes: Export struct return_instance, introduce arch_uretprobe_is_alive() Oleg Nesterov
2015-07-07 12:58   ` Anton Arapov
2015-07-10 11:52   ` Srikar Dronamraju
2015-07-07  1:22 ` [PATCH v2 07/11] uprobes/x86: Reimplement arch_uretprobe_is_alive() Oleg Nesterov
2015-07-07 13:02   ` Anton Arapov
2015-07-10 11:53   ` Srikar Dronamraju
2015-07-07  1:23 ` Oleg Nesterov [this message]
2015-07-07 13:05   ` [PATCH v2 08/11] uprobes: Change handle_trampoline() to flush the frames invalidated by longjmp() Anton Arapov
2015-07-10 11:55   ` Srikar Dronamraju
2015-07-07  1:23 ` [PATCH v2 09/11] uprobes: Change prepare_uretprobe() to (try to) flush the dead frames Oleg Nesterov
2015-07-07 13:07   ` Anton Arapov
2015-07-10 11:57   ` Srikar Dronamraju
2015-07-07  1:23 ` [PATCH v2 10/11] uprobes: Add the "enum rp_check ctx" arg to arch_uretprobe_is_alive() Oleg Nesterov
2015-07-07 13:08   ` Anton Arapov
2015-07-10 12:06   ` Srikar Dronamraju
2015-07-07  1:23 ` [PATCH v2 11/11] uprobes/x86: Make arch_uretprobe_is_alive(RP_CHECK_CALL) more clever Oleg Nesterov
2015-07-07 13:11   ` Anton Arapov
2015-07-10 12:07   ` Srikar Dronamraju
2015-07-10 12:01 ` [PATCH v2 00/11] uprobes: longjmp fixes Pratyush Anand

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=20150707012302.GA7540@redhat.com \
    --to=oleg@redhat.com \
    --cc=ananth@in.ibm.com \
    --cc=arapov@gmail.com \
    --cc=dave.long@linaro.org \
    --cc=dvlasenk@redhat.com \
    --cc=fche@redhat.com \
    --cc=jkenisto@us.ibm.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@kernel.org \
    --cc=mjw@redhat.com \
    --cc=panand@redhat.com \
    --cc=srikar@linux.vnet.ibm.com \
    --cc=willeke@de.ibm.com \
    /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).