mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Steven Rostedt <rostedt@goodmis.org>
To: Peter Zijlstra <peterz@infradead.org>
Cc: tglx@linutronix.de, jpoimboe@redhat.com,
	linux-kernel@vger.kernel.org, x86@kernel.org,
	mhiramat@kernel.org, mbenes@suse.cz, jthierry@redhat.com,
	alexandre.chartre@oracle.com
Subject: Re: [PATCH v5 04/17] x86,ftrace: Fix ftrace_regs_caller() unwind
Date: Tue, 21 Apr 2020 20:33:45 -0400	[thread overview]
Message-ID: <20200421203345.4e165c4b@oasis.local.home> (raw)
In-Reply-To: <20200416115118.749606694@infradead.org>

Hi Peter,

After looking at the code, I realized that the only possible way to do
the "direct call" part, is if the direct function helper is executed
there. All other ftrace_ops, will not call that path.

I added a trampoline that points to ftrace_regs_caller to the direct
ftrace_ops, to force it never to allocate its own trampoline, and thus
no created trampoline should ever do the jump for a direct caller.

By doing this, I was able to move the code around to be a bit simpler,
and not need the double modifications (the double ret).

Of course, if any created trampoline were to touch the ORIG_RAX, then
it would crash. We could always nop that compare in the created
trampoline if that is a concern.

Anyway, I added the below patch on top of your patches and it appears
to work. Thoughts?

-- Steve

diff --git a/arch/x86/kernel/ftrace.c b/arch/x86/kernel/ftrace.c
index 867c126ddabe..2e11250d5647 100644
--- a/arch/x86/kernel/ftrace.c
+++ b/arch/x86/kernel/ftrace.c
@@ -367,13 +367,6 @@ create_trampoline(struct ftrace_ops *ops, unsigned int *tramp_size)
 	if (WARN_ON(ret < 0))
 		goto fail;
 
-	if (ops->flags & FTRACE_OPS_FL_SAVE_REGS) {
-		ip = trampoline + (ftrace_regs_caller_ret - ftrace_regs_caller);
-		ret = probe_kernel_read(ip, (void *)retq, RET_SIZE);
-		if (WARN_ON(ret < 0))
-			goto fail;
-	}
-
 	/*
 	 * The address of the ftrace_ops that is used for this trampoline
 	 * is stored at the end of the trampoline. This will be used to
diff --git a/arch/x86/kernel/ftrace_64.S b/arch/x86/kernel/ftrace_64.S
index 9738ed23964e..1f2afaa8f71f 100644
--- a/arch/x86/kernel/ftrace_64.S
+++ b/arch/x86/kernel/ftrace_64.S
@@ -241,22 +241,9 @@ SYM_INNER_LABEL(ftrace_regs_call, SYM_L_GLOBAL)
 	 */
 	movq ORIG_RAX(%rsp), %rax
 	testq	%rax, %rax
-	jz	1f
+	jnz	1f
 
-	/* Swap the flags with orig_rax */
-	movq MCOUNT_REG_SIZE(%rsp), %rdi
-	movq %rdi, MCOUNT_REG_SIZE-8(%rsp)
-	movq %rax, MCOUNT_REG_SIZE(%rsp)
-
-	restore_mcount_regs 8
-	/* Restore flags */
-	popfq
-
-SYM_INNER_LABEL(ftrace_regs_caller_ret, SYM_L_GLOBAL);
-	UNWIND_HINT_RET_OFFSET
-	jmp	ftrace_epilogue
-
-1:	restore_mcount_regs
+	restore_mcount_regs
 	/* Restore flags */
 	popfq
 
@@ -267,8 +254,19 @@ SYM_INNER_LABEL(ftrace_regs_caller_ret, SYM_L_GLOBAL);
 	 * to the return.
 	 */
 SYM_INNER_LABEL(ftrace_regs_caller_end, SYM_L_GLOBAL)
+	UNWIND_HINT_RET_OFFSET
 	jmp ftrace_epilogue
 
+	/* Swap the flags with orig_rax */
+1:	movq MCOUNT_REG_SIZE(%rsp), %rdi
+	movq %rdi, MCOUNT_REG_SIZE-8(%rsp)
+	movq %rax, MCOUNT_REG_SIZE(%rsp)
+
+	restore_mcount_regs 8
+	/* Restore flags */
+	popfq
+	jmp	ftrace_epilogue
+
 SYM_FUNC_END(ftrace_regs_caller)
 
 
diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c
index 041694a1eb74..8f540eef7511 100644
--- a/kernel/trace/ftrace.c
+++ b/kernel/trace/ftrace.c
@@ -2399,6 +2399,13 @@ struct ftrace_ops direct_ops = {
 	.flags		= FTRACE_OPS_FL_IPMODIFY | FTRACE_OPS_FL_RECURSION_SAFE
 			  | FTRACE_OPS_FL_DIRECT | FTRACE_OPS_FL_SAVE_REGS
 			  | FTRACE_OPS_FL_PERMANENT,
+	/*
+	 * By declaring the main trampoline as this trampoline
+	 * it will never have one allocated for it. This is fine
+	 * as this should only be used if we are in the
+	 * ftrace_ops_list function.
+	 */
+	.trampoline	= FTRACE_REGS_ADDR,
 };
 #endif /* CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS */
 

  parent reply	other threads:[~2020-04-22  0:33 UTC|newest]

Thread overview: 64+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-04-16 11:47 [PATCH v5 00/17] objtool: vmlinux.o and noinstr validation Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 01/17] objtool: Support multiple stack_op per instruction Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 02/17] objtool: Better handle IRET Peter Zijlstra
2020-04-17 11:29   ` Miroslav Benes
2020-04-17 12:25     ` Peter Zijlstra
2020-04-17 12:35       ` Miroslav Benes
2020-04-17 17:37   ` Alexandre Chartre
2020-04-17 18:23     ` Peter Zijlstra
2020-04-17 23:53       ` Andy Lutomirski
2020-04-18 17:18         ` Josh Poimboeuf
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 03/17] objtool: Introduce HINT_RET_OFFSET Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 04/17] x86,ftrace: Fix ftrace_regs_caller() unwind Peter Zijlstra
2020-04-17 19:24   ` Alexandre Chartre
2020-04-22  0:33   ` Steven Rostedt [this message]
2020-04-22  9:44     ` Peter Zijlstra
2020-04-22 13:33       ` Steven Rostedt
2020-04-22 20:20       ` Steven Rostedt
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 05/17] x86,ftrace: Use SIZEOF_PTREGS Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 06/17] x86,ftrace: Shrink ftrace_regs_caller() by one byte Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 07/17] objtool: Remove SAVE/RESTORE hints Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 08/17] objtool: Rename struct cfi_state Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 09/17] objtool: Fix !CFI insn_state propagation Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 10/17] objtool: Implement noinstr validation Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 11/17] objtool: Optimize !vmlinux.o again Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 12/17] objtool: Use sec_offset_hash() for insn_hash Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 13/17] kbuild/objtool: Add objtool-vmlinux.o pass Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 14/17] objtool: Avoid iterating !text section symbols Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 15/17] objtool: Rearrange validate_section() Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 16/17] objtool: Add STT_NOTYPE noinstr validation Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Peter Zijlstra
2020-04-23  7:49   ` tip-bot2 for Peter Zijlstra
2020-04-16 11:47 ` [PATCH v5 17/17] objtool: Also consider .entry.text as noinstr Peter Zijlstra
2020-04-22 22:24   ` [tip: objtool/core] " tip-bot2 for Thomas Gleixner
2020-04-23  7:49   ` tip-bot2 for Thomas Gleixner
2020-04-17 12:33 ` [PATCH v5 00/17] objtool: vmlinux.o and noinstr validation Miroslav Benes
2020-04-17 20:22 ` Alexandre Chartre

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=20200421203345.4e165c4b@oasis.local.home \
    --to=rostedt@goodmis.org \
    --cc=alexandre.chartre@oracle.com \
    --cc=jpoimboe@redhat.com \
    --cc=jthierry@redhat.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mbenes@suse.cz \
    --cc=mhiramat@kernel.org \
    --cc=peterz@infradead.org \
    --cc=tglx@linutronix.de \
    --cc=x86@kernel.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

all inboxes | Powered by JetHome®