From: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
To: Peter Zijlstra <peterz@infradead.org>,
Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>,
Andrew Morton <akpm@linux-foundation.org>,
LKML <linux-kernel@vger.kernel.org>,
Linux-mm <linux-mm@kvack.org>, Ingo Molnar <mingo@elte.hu>,
Andi Kleen <andi@firstfloor.org>,
Christoph Hellwig <hch@infradead.org>,
Steven Rostedt <rostedt@goodmis.org>,
Roland McGrath <roland@hack.frob.com>,
Thomas Gleixner <tglx@linutronix.de>,
Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>,
Arnaldo Carvalho de Melo <acme@infradead.org>,
Anton Arapov <anton@redhat.com>,
Ananth N Mavinakayanahalli <ananth@in.ibm.com>,
Jim Keniston <jkenisto@linux.vnet.ibm.com>,
Stephen Wilson <wilsons@start.ca>
Subject: [PATCH v7 3.2-rc2 27/30] uprobes: x86: introduce xol_was_trapped()
Date: Fri, 18 Nov 2011 16:42:09 +0530 [thread overview]
Message-ID: <20111118111209.10512.44161.sendpatchset@srdronam.in.ibm.com> (raw)
In-Reply-To: <20111118110631.10512.73274.sendpatchset@srdronam.in.ibm.com>
Postpone the signals until we execute the probed insn. This is
simply wrong if xol insn traps and generates the signal itself. Say,
SIGILL/SIGSEGV/etc.
Adds xol_was_trapped() to detect this case. It assumes that anything
like do_page_fault/do_trap/etc sets thread.trap_no != -1.
We add uprobe_task_arch_info->saved_trap_no and change
pre_xol/post_xol to save/restore thread.trap_no, xol_was_trapped()
simply checks that ->trap_no is not equal to UPROBE_TRAP_NO == -1
set by pre_xol().
Original-patch-from: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
---
Changelog (since v6)
- added x86 specific hook for aborting xol.
arch/x86/include/asm/uprobes.h | 7 ++++++-
arch/x86/kernel/uprobes.c | 33 ++++++++++++++++++++++++++++++++-
2 files changed, 38 insertions(+), 2 deletions(-)
diff --git a/arch/x86/include/asm/uprobes.h b/arch/x86/include/asm/uprobes.h
index 99d7d4b..6a47024 100644
--- a/arch/x86/include/asm/uprobes.h
+++ b/arch/x86/include/asm/uprobes.h
@@ -39,16 +39,21 @@ struct uprobe_arch_info {
struct uprobe_task_arch_info {
unsigned long saved_scratch_register;
+ unsigned long saved_trap_no;
};
#else
struct uprobe_arch_info {};
-struct uprobe_task_arch_info {};
+struct uprobe_task_arch_info {
+ unsigned long saved_trap_no;
+};
#endif
struct uprobe;
extern int analyze_insn(struct mm_struct *mm, struct uprobe *uprobe);
extern void set_instruction_pointer(struct pt_regs *regs, unsigned long vaddr);
extern int pre_xol(struct uprobe *uprobe, struct pt_regs *regs);
extern int post_xol(struct uprobe *uprobe, struct pt_regs *regs);
+extern bool xol_was_trapped(struct task_struct *tsk);
extern int uprobe_exception_notify(struct notifier_block *self,
unsigned long val, void *data);
+extern void abort_xol(struct pt_regs *regs, struct uprobe *uprobe);
#endif /* _ASM_UPROBES_H */
diff --git a/arch/x86/kernel/uprobes.c b/arch/x86/kernel/uprobes.c
index 0792fc8..3f0eb4e 100644
--- a/arch/x86/kernel/uprobes.c
+++ b/arch/x86/kernel/uprobes.c
@@ -409,6 +409,8 @@ void set_instruction_pointer(struct pt_regs *regs, unsigned long vaddr)
regs->ip = vaddr;
}
+#define UPROBE_TRAP_NO -1ul
+
/*
* pre_xol - prepare to execute out of line.
* @uprobe: the probepoint information.
@@ -424,6 +426,9 @@ int pre_xol(struct uprobe *uprobe, struct pt_regs *regs)
{
struct uprobe_task_arch_info *tskinfo = ¤t->utask->tskinfo;
+ tskinfo->saved_trap_no = current->thread.trap_no;
+ current->thread.trap_no = UPROBE_TRAP_NO;
+
regs->ip = current->utask->xol_vaddr;
if (uprobe->fixups & UPROBES_FIX_RIP_AX) {
tskinfo->saved_scratch_register = regs->ax;
@@ -439,6 +444,11 @@ int pre_xol(struct uprobe *uprobe, struct pt_regs *regs)
#else
int pre_xol(struct uprobe *uprobe, struct pt_regs *regs)
{
+ struct uprobe_task_arch_info *tskinfo = ¤t->utask->tskinfo;
+
+ tskinfo->saved_trap_no = current->thread.trap_no;
+ current->thread.trap_no = UPROBE_TRAP_NO;
+
regs->ip = current->utask->xol_vaddr;
return 0;
}
@@ -494,7 +504,8 @@ static void handle_riprel_post_xol(struct uprobe *uprobe,
* Fall through to handle stuff like "jmpq *...(%rip)" and
* "callq *...(%rip)".
*/
- *correction += 4;
+ if (correction)
+ *correction += 4;
}
}
#else
@@ -504,6 +515,14 @@ static void handle_riprel_post_xol(struct uprobe *uprobe,
}
#endif
+bool xol_was_trapped(struct task_struct *tsk)
+{
+ if (tsk->thread.trap_no != UPROBE_TRAP_NO)
+ return true;
+
+ return false;
+}
+
/*
* Called after single-stepping. To avoid the SMP problems that can
* occur when we temporarily put back the original opcode to
@@ -534,6 +553,9 @@ int post_xol(struct uprobe *uprobe, struct pt_regs *regs)
int result = 0;
long correction;
+ WARN_ON_ONCE(current->thread.trap_no != UPROBE_TRAP_NO);
+
+ current->thread.trap_no = utask->tskinfo.saved_trap_no;
correction = (long)(utask->vaddr - utask->xol_vaddr);
handle_riprel_post_xol(uprobe, regs, &correction);
if (uprobe->fixups & UPROBES_FIX_IP)
@@ -571,3 +593,12 @@ int uprobe_exception_notify(struct notifier_block *self,
}
return ret;
}
+
+void abort_xol(struct pt_regs *regs, struct uprobe *uprobe)
+{
+ struct uprobe_task *utask = current->utask;
+
+ current->thread.trap_no = utask->tskinfo.saved_trap_no;
+ handle_riprel_post_xol(uprobe, regs, NULL);
+ set_instruction_pointer(regs, utask->vaddr);
+}
next prev parent reply other threads:[~2011-11-18 11:38 UTC|newest]
Thread overview: 106+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-11-18 11:06 [PATCH v7 3.2-rc2 0/30] uprobes patchset with perf probe support Srikar Dronamraju
2011-11-18 11:06 ` [PATCH v7 3.2-rc2 1/30] uprobes: Auxillary routines to insert, find, delete uprobes Srikar Dronamraju
2011-11-23 18:23 ` Peter Zijlstra
2011-11-18 11:07 ` [PATCH v7 3.2-rc2 2/30] uprobes: Allow multiple consumers for an uprobe Srikar Dronamraju
2011-11-18 11:07 ` [PATCH v7 3.2-rc2 3/30] uprobes: register/unregister probes Srikar Dronamraju
2011-11-23 16:09 ` Peter Zijlstra
2011-11-23 16:11 ` Peter Zijlstra
2011-11-24 14:39 ` Srikar Dronamraju
2011-11-23 16:22 ` Peter Zijlstra
2011-11-23 16:27 ` Peter Zijlstra
2011-11-23 16:35 ` Peter Zijlstra
2011-11-28 15:29 ` Peter Zijlstra
2011-11-29 7:48 ` Srikar Dronamraju
2011-11-29 10:52 ` Peter Zijlstra
2011-12-01 13:41 ` Srikar Dronamraju
2011-12-01 13:20 ` Peter Zijlstra
2011-11-18 11:07 ` [PATCH v7 3.2-rc2 4/30] uprobes: Define hooks for mmap/munmap Srikar Dronamraju
2011-11-23 17:13 ` Peter Zijlstra
2011-11-23 18:10 ` Peter Zijlstra
2011-11-24 13:47 ` Srikar Dronamraju
2011-11-24 14:13 ` Peter Zijlstra
2011-11-24 14:25 ` Srikar Dronamraju
2011-11-28 14:59 ` Peter Zijlstra
2011-11-29 8:33 ` Srikar Dronamraju
2011-11-29 11:48 ` Peter Zijlstra
2011-11-29 15:05 ` Peter Zijlstra
2011-11-30 5:50 ` Srikar Dronamraju
2011-11-29 16:22 ` Srikar Dronamraju
2011-11-30 12:25 ` Peter Zijlstra
2011-12-01 5:40 ` Srikar Dronamraju
2011-12-01 11:36 ` Peter Zijlstra
2011-12-01 13:24 ` Srikar Dronamraju
2011-11-30 5:30 ` Srikar Dronamraju
2011-11-23 18:15 ` Peter Zijlstra
2011-11-23 19:50 ` Steven Rostedt
2011-11-24 13:37 ` Srikar Dronamraju
2011-11-24 13:47 ` Peter Zijlstra
2011-11-18 11:07 ` [PATCH v7 3.2-rc2 5/30] uprobes: copy of the original instruction Srikar Dronamraju
2011-11-23 18:26 ` Peter Zijlstra
2011-11-23 18:40 ` Peter Zijlstra
2011-11-23 19:49 ` Steven Rostedt
2011-11-23 20:52 ` Peter Zijlstra
2011-11-24 12:50 ` Srikar Dronamraju
2011-11-28 14:23 ` Peter Zijlstra
2011-11-18 11:07 ` [PATCH v7 3.2-rc2 6/30] uprobes: define fixups Srikar Dronamraju
2011-11-18 11:07 ` [PATCH v7 3.2-rc2 7/30] uprobes: uprobes arch info Srikar Dronamraju
2011-11-18 11:08 ` [PATCH v7 3.2-rc2 8/30] x86: analyze instruction and determine fixups Srikar Dronamraju
2011-11-30 18:57 ` Oleg Nesterov
2011-12-01 5:52 ` Srikar Dronamraju
2011-11-18 11:08 ` [PATCH v7 3.2-rc2 9/30] uprobes: Background page replacement Srikar Dronamraju
2011-11-25 14:29 ` Peter Zijlstra
2011-11-25 14:54 ` Peter Zijlstra
2011-11-26 2:25 ` Srikar Dronamraju
2011-11-28 14:13 ` Peter Zijlstra
2011-11-29 7:49 ` Srikar Dronamraju
2011-11-28 15:01 ` Peter Zijlstra
2011-11-18 11:08 ` [PATCH v7 3.2-rc2 10/30] x86: Set instruction pointer Srikar Dronamraju
2011-11-18 11:08 ` [PATCH v7 3.2-rc2 11/30] x86: Introduce TIF_UPROBE FLAG Srikar Dronamraju
2011-11-18 11:09 ` [PATCH v7 3.2-rc2 12/30] uprobes: Handle breakpoint and Singlestep Srikar Dronamraju
2011-11-25 15:24 ` Peter Zijlstra
2011-11-26 2:22 ` Srikar Dronamraju
2011-11-18 11:09 ` [PATCH v7 3.2-rc2 13/30] x86: define a x86 specific exception notifier Srikar Dronamraju
2011-11-18 11:09 ` [PATCH v7 3.2-rc2 14/30] uprobe: register " Srikar Dronamraju
2011-11-18 11:09 ` [PATCH v7 3.2-rc2 15/30] x86: Define x86_64 specific uprobe_task_arch_info structure Srikar Dronamraju
2011-11-18 11:09 ` [PATCH v7 3.2-rc2 16/30] uprobes: Introduce " Srikar Dronamraju
2011-11-18 11:09 ` [PATCH v7 3.2-rc2 17/30] x86: arch specific hooks for pre/post singlestep handling Srikar Dronamraju
2011-11-18 11:10 ` [PATCH v7 3.2-rc2 18/30] uprobes: slot allocation Srikar Dronamraju
2011-11-18 11:10 ` [PATCH v7 3.2-rc2 19/30] tracing: modify is_delete, is_return from ints to bool Srikar Dronamraju
2011-11-23 19:24 ` Steven Rostedt
2011-11-18 11:10 ` [PATCH v7 3.2-rc2 20/30] tracing: Extract out common code for kprobes/uprobes traceevents Srikar Dronamraju
2011-11-23 19:32 ` Steven Rostedt
2011-11-24 13:12 ` Srikar Dronamraju
2011-11-18 11:10 ` [PATCH v7 3.2-rc2 21/30] tracing: uprobes trace_event interface Srikar Dronamraju
2011-11-18 11:10 ` [PATCH v7 3.2-rc2 22/30] perf: rename target_module to target Srikar Dronamraju
2011-11-18 11:11 ` [PATCH v7 3.2-rc2 23/30] perf: perf interface for uprobes Srikar Dronamraju
2011-11-18 11:11 ` [PATCH v7 3.2-rc2 24/30] perf: show possible probes in a given executable file or library Srikar Dronamraju
2011-11-18 11:11 ` [PATCH v7 3.2-rc2 25/30] uprobes: call post_xol() unconditionally Srikar Dronamraju
2011-11-18 11:11 ` [PATCH v7 3.2-rc2 26/30] uprobes: introduce uprobe_deny_signal() Srikar Dronamraju
2011-11-18 11:12 ` Srikar Dronamraju [this message]
2011-11-18 11:12 ` [PATCH v7 3.2-rc2 28/30] uprobes: introduce UTASK_SSTEP_TRAPPED logic Srikar Dronamraju
2011-11-18 11:12 ` [PATCH v7 3.2-rc2 29/30] uprobes: Introduce uprobe flags Srikar Dronamraju
2011-11-18 11:12 ` [PATCH v7 3.2-rc2 30/30] x86: skip singlestep where possible Srikar Dronamraju
2011-11-22 5:03 ` [PATCH v7 3.2-rc2 0/30] uprobes patchset with perf probe support Srikar Dronamraju
2011-11-22 14:49 ` Stephen Rothwell
2011-11-23 13:20 ` Srikar Dronamraju
2011-11-23 13:38 ` Stephen Rothwell
2011-11-28 19:06 ` [PATCH RFC 0/5] uprobes: kill xol vma Oleg Nesterov
2011-11-28 19:06 ` [PATCH 1/5] uprobes: kill pre_ssout(), introduce set_xol_ip() Oleg Nesterov
2011-11-28 19:06 ` [PATCH 2/5] uprobes: introduce uprobe_switch_to() Oleg Nesterov
2011-11-28 19:53 ` Peter Zijlstra
2011-11-29 17:18 ` Oleg Nesterov
2011-11-30 12:11 ` Peter Zijlstra
2011-11-30 17:10 ` Oleg Nesterov
2011-11-28 19:07 ` [PATCH 3/5] uprobes: introduce uprobe_xol_slots[NR_CPUS] Oleg Nesterov
2011-11-28 19:48 ` Peter Zijlstra
2011-11-28 19:52 ` Peter Zijlstra
2011-11-29 18:24 ` Oleg Nesterov
2011-11-28 19:07 ` [PATCH 4/5] uprobes: teach set_xol_ip() to use uprobe_xol_slots[] Oleg Nesterov
2011-11-28 19:07 ` [PATCH 5/5] uprobes: remove the uprobes_xol_area code Oleg Nesterov
2011-11-28 19:57 ` [PATCH RFC 0/5] uprobes: kill xol vma Peter Zijlstra
2011-11-29 10:30 ` Srikar Dronamraju
2011-11-29 18:26 ` Oleg Nesterov
2011-11-30 16:15 ` Andi Kleen
2011-11-30 16:20 ` Peter Zijlstra
2011-11-30 18:47 ` Oleg Nesterov
2011-12-12 17:30 ` Oleg Nesterov
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=20111118111209.10512.44161.sendpatchset@srdronam.in.ibm.com \
--to=srikar@linux.vnet.ibm.com \
--cc=acme@infradead.org \
--cc=akpm@linux-foundation.org \
--cc=ananth@in.ibm.com \
--cc=andi@firstfloor.org \
--cc=anton@redhat.com \
--cc=hch@infradead.org \
--cc=jkenisto@linux.vnet.ibm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=masami.hiramatsu.pt@hitachi.com \
--cc=mingo@elte.hu \
--cc=oleg@redhat.com \
--cc=peterz@infradead.org \
--cc=roland@hack.frob.com \
--cc=rostedt@goodmis.org \
--cc=tglx@linutronix.de \
--cc=torvalds@linux-foundation.org \
--cc=wilsons@start.ca \
/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).