Skip to contentSkip to conversation

fprobe: Introduce fprobe function entry/exit probe

PATCH In review Masami Hiramatsu 4 participants 13 messages Mar 23, 2022 – Mar 25, 2022

Subsystems (5)
Networking Network drivers x86 BPF +1 more subsystems
lore
Share & tools
lore
#0 lore raw

v13 cover fprobe: Introduce fprobe function entry/exit probe

Hi,

Here is the 13th version of rethook x86 port. This is developed for a part of fprobe series [1] for hooking function return. But since I forgot to send it to arch maintainers, that caused conflict with IBT and SLS mitigation series. Now I picked the x86 rethook part and send it to x86 maintainers to be reviewed.

[1] https://lore.kernel.org/all/164735281449.1084943.12438881786173547153.stgit@devnote2/T/#u

Note that this patch is still for the bpf-next since the rethook itself is on the bpf-next tree. But since this also uses the ANNOTATE_NOENDBR macro which has been introduced by IBT/ENDBR patch, to build this series you need to merge the tip/master branch with the bpf-next. (hopefully, it is rebased soon)

The fprobe itself is for providing the function entry/exit probe with multiple probe point. The rethook is a sub-feature to hook the function return as same as kretprobe does. Eventually, I would like to replace the kretprobe's trampoline with this rethook.

Thank you,

---

Masami Hiramatsu (1):
rethook: x86: Add rethook x86 implementation

 arch/x86/Kconfig                 |    1 
 arch/x86/include/asm/unwind.h    |    8 ++-
 arch/x86/kernel/Makefile         |    1 
 arch/x86/kernel/kprobes/common.h |    1 
 arch/x86/kernel/rethook.c        |  121 ++++++++++++++++++++++++++++++++++++++
 5 files changed, 131 insertions(+), 1 deletion(-)
 create mode 100644 arch/x86/kernel/rethook.c

--
Masami Hiramatsu (Linaro) <[email protected]>

#1 lore raw

PATCH v13 1/1 rethook: x86: Add rethook x86 implementation

Add rethook for x86 implementation. Most of the code has been copied from kretprobes on x86.

---
Changes in v13:
- Add no ENDBR annotation for return trampoline code.
- Replace ret with ASM_RET for Straight-Line-Speculation mitigation.
Changes in v11:
- Mark arch_rethook_fixup_return() as NOKPROBE_SYMBOL.
- Add a function prototype of arch_rethook_trampoline_callback to suppress warning.
Changes in v10:
- Add a dummy @mcount to arch_rethook_prepare().
Changes in v5:
- Fix a build error if !CONFIG_KRETPROBES and !CONFIG_RETHOOK.
Changes in v4:
- fix stack backtrace as same as kretprobe does.

---
 arch/x86/Kconfig                 |    1 
 arch/x86/include/asm/unwind.h    |    8 ++-
 arch/x86/kernel/Makefile         |    1 
 arch/x86/kernel/kprobes/common.h |    1 
 arch/x86/kernel/rethook.c        |  121 ++++++++++++++++++++++++++++++++++++++
 5 files changed, 131 insertions(+), 1 deletion(-)
 create mode 100644 arch/x86/kernel/rethook.c
arch/x86/Kconfig +1 -0
index 9b356da6f46b..1270b65a5546 100644
@@ -221,6 +221,7 @@ config X86
 	select HAVE_KPROBES_ON_FTRACE
 	select HAVE_FUNCTION_ERROR_INJECTION
 	select HAVE_KRETPROBES
+	select HAVE_RETHOOK
 	select HAVE_KVM
 	select HAVE_LIVEPATCH			if X86_64
 	select HAVE_MIXED_BREAKPOINTS_REGS
arch/x86/include/asm/unwind.h +7 -1
index 2a1f8734416d..192df5b2094d 100644
@@ -5,6 +5,7 @@
 #include <linux/sched.h>
 #include <linux/ftrace.h>
 #include <linux/kprobes.h>
+#include <linux/rethook.h>
 #include <asm/ptrace.h>
 #include <asm/stacktrace.h>
 
@@ -16,7 +17,7 @@ struct unwind_state {
 	unsigned long stack_mask;
 	struct task_struct *task;
 	int graph_idx;
-#ifdef CONFIG_KRETPROBES
+#if defined(CONFIG_KRETPROBES) || defined(CONFIG_RETHOOK)
 	struct llist_node *kr_cur;
 #endif
 	bool error;
@@ -107,6 +108,11 @@ static inline
 unsigned long unwind_recover_kretprobe(struct unwind_state *state,
 				       unsigned long addr, unsigned long *addr_p)
 {
+#ifdef CONFIG_RETHOOK
+	if (is_rethook_trampoline(addr))
+		return rethook_find_ret_addr(state->task, (unsigned long)addr_p,
+					     &state->kr_cur);
+#endif
 #ifdef CONFIG_KRETPROBES
 	return is_kretprobe_trampoline(addr) ?
 		kretprobe_find_ret_addr(state->task, addr_p, &state->kr_cur) :
arch/x86/kernel/Makefile +1 -0
index 6aef9ee28a39..792a893a5cc5 100644
@@ -106,6 +106,7 @@ obj-$(CONFIG_FUNCTION_GRAPH_TRACER) += ftrace.o
 obj-$(CONFIG_FTRACE_SYSCALLS)	+= ftrace.o
 obj-$(CONFIG_X86_TSC)		+= trace_clock.o
 obj-$(CONFIG_TRACING)		+= trace.o
+obj-$(CONFIG_RETHOOK)		+= rethook.o
 obj-$(CONFIG_CRASH_CORE)	+= crash_core_$(BITS).o
 obj-$(CONFIG_KEXEC_CORE)	+= machine_kexec_$(BITS).o
 obj-$(CONFIG_KEXEC_CORE)	+= relocate_kernel_$(BITS).o crash.o
arch/x86/kernel/kprobes/common.h +1 -0
index 7d3a2e2daf01..c993521d4933 100644
@@ -6,6 +6,7 @@
 
 #include <asm/asm.h>
 #include <asm/frame.h>
+#include <asm/insn.h>
 
 #ifdef CONFIG_X86_64
 
arch/x86/kernel/rethook.c new file +121 -0

121 lines not on this page · raw message

#9 lore raw
In reply to Masami Hiramatsu

Hi,

The title is not updated. It should be;

rethook: x86: Add rethook x86 porting (drived from "fprobe: Introduce fprobe function entry/exit probe" series)

Thank you,

Quoted 39 lines · On Wed, 23 Mar 2022 11:34:46 +0900 Masami Hiramatsu <[email protected]> wrote:

Hi,

Here is the 13th version of rethook x86 port. This is developed for a part of fprobe series [1] for hooking function return. But since I forgot to send it to arch maintainers, that caused conflict with IBT and SLS mitigation series. Now I picked the x86 rethook part and send it to x86 maintainers to be reviewed.

[1] https://lore.kernel.org/all/164735281449.1084943.12438881786173547153.stgit@devnote2/T/#u

Note that this patch is still for the bpf-next since the rethook itself is on the bpf-next tree. But since this also uses the ANNOTATE_NOENDBR macro which has been introduced by IBT/ENDBR patch, to build this series you need to merge the tip/master branch with the bpf-next. (hopefully, it is rebased soon)

The fprobe itself is for providing the function entry/exit probe with multiple probe point. The rethook is a sub-feature to hook the function return as same as kretprobe does. Eventually, I would like to replace the kretprobe's trampoline with this rethook.

Thank you,

---

Masami Hiramatsu (1):
rethook: x86: Add rethook x86 implementation

 arch/x86/Kconfig                 |    1 
 arch/x86/include/asm/unwind.h    |    8 ++-
 arch/x86/kernel/Makefile         |    1 
 arch/x86/kernel/kprobes/common.h |    1 
 arch/x86/kernel/rethook.c        |  121 ++++++++++++++++++++++++++++++++++++++
 5 files changed, 131 insertions(+), 1 deletion(-)
 create mode 100644 arch/x86/kernel/rethook.c

--
Masami Hiramatsu (Linaro) <[email protected]>

Signature · 1 line
Masami Hiramatsu <[email protected]>
maintainer
#2 lore raw

Re: [PATCH v13 bpf-next 1/1] rethook: x86: Add rethook x86 implementation

In reply to Masami Hiramatsu
Quoted 2 lines · Masami Hiramatsu

Add rethook for x86 implementation. Most of the code has been copied from kretprobes on x86.

Right; as said, I'm really unhappy with growing a carbon copy of this stuff instead of sharing. Can we *please* keep it a single instance? Them being basically indentical, it should be trivial to have CONFIG_KPROBE_ON_RETHOOK (or somesuch) and just share this.

Also, what's rethook for anyway?

Quoted 35 lines
arch/x86/kernel/kprobes/common.h +1 -0
index 7d3a2e2daf01..c993521d4933 100644
@@ -6,6 +6,7 @@
 
 #include <asm/asm.h>
 #include <asm/frame.h>
+#include <asm/insn.h>
 
 #ifdef CONFIG_X86_64
 
arch/x86/kernel/rethook.c new file +17 -0
new file mode 100644
index 000000000000..3e916361c33b
@@ -0,0 +1,121 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * x86 implementation of rethook. Mostly copied from arch/x86/kernel/kprobes/core.c.
+ */
+#include <linux/bug.h>
+#include <linux/rethook.h>
+#include <linux/kprobes.h>
+#include <linux/objtool.h>
+
+#include "kprobes/common.h"
+
+__visible void arch_rethook_trampoline_callback(struct pt_regs *regs);
+
+/*
+ * When a target function returns, this code saves registers and calls
+ * arch_rethook_trampoline_callback(), which calls the rethook handler.
+ */

arch/x86/kernel/rethook.c line 27

Quoted 10 lines
arch/x86/kernel/rethook.c new file +10 -0
+asm(
+	".text\n"
+	".global arch_rethook_trampoline\n"
+	".type arch_rethook_trampoline, @function\n"
+	"arch_rethook_trampoline:\n"
+#ifdef CONFIG_X86_64
+	ANNOTATE_NOENDBR	/* This is only jumped from ret instruction */
+	/* Push a fake return address to tell the unwinder it's a kretprobe. */
+	"	pushq $arch_rethook_trampoline\n"
+	UNWIND_HINT_FUNC
	"	pushq $" __stringify(__KERNEL_DS) "\n" /* %ss */
	/* Save the 'sp - 16', this will be fixed later. */

arch/x86/kernel/rethook.c line 34

Quoted 6 lines
arch/x86/kernel/rethook.c new file +6 -0
+	"	pushq %rsp\n"
+	"	pushfq\n"
+	SAVE_REGS_STRING
+	"	movq %rsp, %rdi\n"
+	"	call arch_rethook_trampoline_callback\n"
+	RESTORE_REGS_STRING

/* In the callback function, 'regs->flags' is copied to 'regs->ss'. */

this comment could do with a 'why' though... Because neither this nor the one in the handler really explains why it is important to have popf last

" addq $16, %rsp\n"

arch/x86/kernel/rethook.c line 38

Quoted 2 lines
arch/x86/kernel/rethook.c new file +2 -0
+	"	popfq\n"
+#else

same for i386:

arch/x86/kernel/rethook.c line 41

Quoted 3 lines
arch/x86/kernel/rethook.c new file +3 -0
+	/* Push a fake return address to tell the unwinder it's a kretprobe. */
+	"	pushl $arch_rethook_trampoline\n"
+	UNWIND_HINT_FUNC
	/* Save the 'sp - 8', this will be fixed later. */
	"	pushl %ss\n"

arch/x86/kernel/rethook.c line 48

Quoted 6 lines
arch/x86/kernel/rethook.c new file +6 -0
+	"	pushl %esp\n"
+	"	pushfl\n"
+	SAVE_REGS_STRING
+	"	movl %esp, %eax\n"
+	"	call arch_rethook_trampoline_callback\n"
+	RESTORE_REGS_STRING
	/* In the callback function, 'regs->flags' is copied to 'regs->ss'. */
	"	addl $8, %esp\n"
Quoted 11 lines
arch/x86/kernel/rethook.c new file +11 -0
+	"	popfl\n"
+#endif
+	ASM_RET
+	".size arch_rethook_trampoline, .-arch_rethook_trampoline\n"
+);
+NOKPROBE_SYMBOL(arch_rethook_trampoline);
+
+/*
+ * Called from arch_rethook_trampoline
+ */
+__used __visible void arch_rethook_trampoline_callback(struct pt_regs *regs)

arch/x86/kernel/rethook.c line 71

Quoted 10 lines
arch/x86/kernel/rethook.c new file +10 -0
+{
+	unsigned long *frame_pointer;
+
+	/* fixup registers */
+	regs->cs = __KERNEL_CS;
+#ifdef CONFIG_X86_32
+	regs->gs = 0;
+#endif
+	regs->ip = (unsigned long)&arch_rethook_trampoline;
+	regs->orig_ax = ~0UL;
	regs->sp += 2*sizeof(long);

arch/x86/kernel/rethook.c line 85

Quoted 13 lines
arch/x86/kernel/rethook.c new file +13 -0
+	frame_pointer = &regs->sp + 1;
+
+	/*
+	 * The return address at 'frame_pointer' is recovered by the
+	 * arch_rethook_fixup_return() which called from this
+	 * rethook_trampoline_handler().
+	 */
+	rethook_trampoline_handler(regs, (unsigned long)frame_pointer);
+
+	/*
+	 * Copy FLAGS to 'pt_regs::sp' so that arch_rethook_trapmoline()
+	 * can do RET right after POPF.
+	 */
	regs->ss = regs->flags;
Quoted 2 lines
arch/x86/kernel/rethook.c new file +2 -0
+}
+NOKPROBE_SYMBOL(arch_rethook_trampoline_callback);
#3 lore raw

Re: [PATCH v13 bpf-next 1/1] rethook: x86: Add rethook x86 implementation

Quoted 6 lines · On Wed, 23 Mar 2022 09:05:26 +0100 Peter Zijlstra <[email protected]> wrote:
Quoted 2 lines · Masami Hiramatsu

Add rethook for x86 implementation. Most of the code has been copied from kretprobes on x86.

Right; as said, I'm really unhappy with growing a carbon copy of this stuff instead of sharing. Can we *please* keep it a single instance?

OK, then let me update the kprobe side too.

Quoted 2 lines

Them being basically indentical, it should be trivial to have CONFIG_KPROBE_ON_RETHOOK (or somesuch) and just share this.

Yes, ideally it should use CONFIG_HAVE_RETHOOK since the rethook arch port must be a copy of the kretprobe implementation. But for safety, I think having CONFIG_KPROBE_ON_RETHOOK is a good idea until replacing all kretprobe implementations.

Quoted 2 lines

Also, what's rethook for anyway?

Rethook is a feature which hooks the function return. Most of the logic came from the kretprobe. Simply to say, 'kretprobe - kprobe' is the rethook :)

Thank you,

Quoted 117 lines
Quoted 36 lines
arch/x86/kernel/kprobes/common.h +1 -0
index 7d3a2e2daf01..c993521d4933 100644
@@ -6,6 +6,7 @@
 
 #include <asm/asm.h>
 #include <asm/frame.h>
+#include <asm/insn.h>
 
 #ifdef CONFIG_X86_64
 
arch/x86/kernel/rethook.c new file +17 -0
new file mode 100644
index 000000000000..3e916361c33b
@@ -0,0 +1,121 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * x86 implementation of rethook. Mostly copied from arch/x86/kernel/kprobes/core.c.
+ */
+#include <linux/bug.h>
+#include <linux/rethook.h>
+#include <linux/kprobes.h>
+#include <linux/objtool.h>
+
+#include "kprobes/common.h"
+
+__visible void arch_rethook_trampoline_callback(struct pt_regs *regs);
+
+/*
+ * When a target function returns, this code saves registers and calls
+ * arch_rethook_trampoline_callback(), which calls the rethook handler.
+ */

arch/x86/kernel/rethook.c line 27

Quoted 10 lines
arch/x86/kernel/rethook.c new file +10 -0
+asm(
+	".text\n"
+	".global arch_rethook_trampoline\n"
+	".type arch_rethook_trampoline, @function\n"
+	"arch_rethook_trampoline:\n"
+#ifdef CONFIG_X86_64
+	ANNOTATE_NOENDBR	/* This is only jumped from ret instruction */
+	/* Push a fake return address to tell the unwinder it's a kretprobe. */
+	"	pushq $arch_rethook_trampoline\n"
+	UNWIND_HINT_FUNC
	"	pushq $" __stringify(__KERNEL_DS) "\n" /* %ss */
	/* Save the 'sp - 16', this will be fixed later. */

arch/x86/kernel/rethook.c line 34

Quoted 6 lines
arch/x86/kernel/rethook.c new file +6 -0
+	"	pushq %rsp\n"
+	"	pushfq\n"
+	SAVE_REGS_STRING
+	"	movq %rsp, %rdi\n"
+	"	call arch_rethook_trampoline_callback\n"
+	RESTORE_REGS_STRING

/* In the callback function, 'regs->flags' is copied to 'regs->ss'. */

this comment could do with a 'why' though... Because neither this nor the one in the handler really explains why it is important to have popf last

" addq $16, %rsp\n"

arch/x86/kernel/rethook.c line 38

Quoted 2 lines
arch/x86/kernel/rethook.c new file +2 -0
+	"	popfq\n"
+#else

same for i386:

arch/x86/kernel/rethook.c line 41

Quoted 3 lines
arch/x86/kernel/rethook.c new file +3 -0
+	/* Push a fake return address to tell the unwinder it's a kretprobe. */
+	"	pushl $arch_rethook_trampoline\n"
+	UNWIND_HINT_FUNC
	/* Save the 'sp - 8', this will be fixed later. */
	"	pushl %ss\n"

arch/x86/kernel/rethook.c line 48

Quoted 6 lines
arch/x86/kernel/rethook.c new file +6 -0
+	"	pushl %esp\n"
+	"	pushfl\n"
+	SAVE_REGS_STRING
+	"	movl %esp, %eax\n"
+	"	call arch_rethook_trampoline_callback\n"
+	RESTORE_REGS_STRING
	/* In the callback function, 'regs->flags' is copied to 'regs->ss'. */
	"	addl $8, %esp\n"
Quoted 11 lines
arch/x86/kernel/rethook.c new file +11 -0
+	"	popfl\n"
+#endif
+	ASM_RET
+	".size arch_rethook_trampoline, .-arch_rethook_trampoline\n"
+);
+NOKPROBE_SYMBOL(arch_rethook_trampoline);
+
+/*
+ * Called from arch_rethook_trampoline
+ */
+__used __visible void arch_rethook_trampoline_callback(struct pt_regs *regs)

arch/x86/kernel/rethook.c line 71

Quoted 10 lines
arch/x86/kernel/rethook.c new file +10 -0
+{
+	unsigned long *frame_pointer;
+
+	/* fixup registers */
+	regs->cs = __KERNEL_CS;
+#ifdef CONFIG_X86_32
+	regs->gs = 0;
+#endif
+	regs->ip = (unsigned long)&arch_rethook_trampoline;
+	regs->orig_ax = ~0UL;
	regs->sp += 2*sizeof(long);

arch/x86/kernel/rethook.c line 85

Quoted 13 lines
arch/x86/kernel/rethook.c new file +13 -0
+	frame_pointer = &regs->sp + 1;
+
+	/*
+	 * The return address at 'frame_pointer' is recovered by the
+	 * arch_rethook_fixup_return() which called from this
+	 * rethook_trampoline_handler().
+	 */
+	rethook_trampoline_handler(regs, (unsigned long)frame_pointer);
+
+	/*
+	 * Copy FLAGS to 'pt_regs::sp' so that arch_rethook_trapmoline()
+	 * can do RET right after POPF.
+	 */
	regs->ss = regs->flags;
Quoted 2 lines
arch/x86/kernel/rethook.c new file +2 -0
+}
+NOKPROBE_SYMBOL(arch_rethook_trampoline_callback);
Signature · 1 line
Masami Hiramatsu <[email protected]>
maintainer
#4 lore raw

Re: [PATCH v13 bpf-next 1/1] rethook: x86: Add rethook x86 implementation

Quoted 5 lines · Masami Hiramatsu
Quoted 1 line

Also, what's rethook for anyway?

Rethook is a feature which hooks the function return. Most of the logic came from the kretprobe. Simply to say, 'kretprobe - kprobe' is the rethook :)

I got that far, but why did you take the bother to do these patches? Why wasn't 'use kretprobe' a good enough option?

#10 lore raw
In reply to Masami Hiramatsu
Quoted 1 line · Masami Hiramatsu

Hi,

Hi Masami,

Quoted 7 lines

Here is the 13th version of rethook x86 port. This is developed for a part of fprobe series [1] for hooking function return. But since I forgot to send it to arch maintainers, that caused conflict with IBT and SLS mitigation series. Now I picked the x86 rethook part and send it to x86 maintainers to be reviewed.

[1] https://lore.kernel.org/all/164735281449.1084943.12438881786173547153.stgit@devnote2/T/#u

As mentioned elsewhere, I have similar (though not identical) concerns to Peter for the arm64 patch, which was equally unreviewed by maintainers, and the overall structure.

Quoted 5 lines

Note that this patch is still for the bpf-next since the rethook itself is on the bpf-next tree. But since this also uses the ANNOTATE_NOENDBR macro which has been introduced by IBT/ENDBR patch, to build this series you need to merge the tip/master branch with the bpf-next. (hopefully, it is rebased soon)

I thought we were going to drop the series from the bpf-next tree so that this could all go through review it had missed thusfar.

Is that still the plan? What's going on?

Quoted 4 lines

The fprobe itself is for providing the function entry/exit probe with multiple probe point. The rethook is a sub-feature to hook the function return as same as kretprobe does. Eventually, I would like to replace the kretprobe's trampoline with this rethook.

Can we please start by converting each architecture to rethook?

Ideally we'd unify things such that each architecture only needs *one* return trampoline that both ftrace and krpboes can use, which'd be significantly easier to get right and manage.

Thanks,
Mark.

#11 lore raw
Quoted 11 lines · On Wed, 23 Mar 2022 14:18:40 +0000 Mark Rutland <[email protected]> wrote:
Quoted 1 line · Masami Hiramatsu

Hi,

Hi Masami,

Quoted 6 lines

Here is the 13th version of rethook x86 port. This is developed for a part of fprobe series [1] for hooking function return. But since I forgot to send it to arch maintainers, that caused conflict with IBT and SLS mitigation series. Now I picked the x86 rethook part and send it to x86 maintainers to be reviewed.

Quoted 5 lines
Quoted 1 line

[1] https://lore.kernel.org/all/164735281449.1084943.12438881786173547153.stgit@devnote2/T/#u

As mentioned elsewhere, I have similar (though not identical) concerns to Peter for the arm64 patch, which was equally unreviewed by maintainers, and the overall structure.

Yes, those should be reviewed by arch maintainers.

Quoted 11 lines
Quoted 6 lines

Note that this patch is still for the bpf-next since the rethook itself is on the bpf-next tree. But since this also uses the ANNOTATE_NOENDBR macro which has been introduced by IBT/ENDBR patch, to build this series you need to merge the tip/master branch with the bpf-next. (hopefully, it is rebased soon)

I thought we were going to drop the series from the bpf-next tree so that this could all go through review it had missed thusfar.

Is that still the plan? What's going on?

Now the arm64 (and other arch) port is reverted from bpf-next. I'll send those to you soon.
Since bpf-next is focusing on x86 at first, I chose this for review in this version. Sorry for confusion.

Quoted 7 lines
Quoted 5 lines

The fprobe itself is for providing the function entry/exit probe with multiple probe point. The rethook is a sub-feature to hook the function return as same as kretprobe does. Eventually, I would like to replace the kretprobe's trampoline with this rethook.

Can we please start by converting each architecture to rethook?

Yes. As Peter pointed, I'm planning to add a kretprobe patches to use rethook if available in that series. let me prepare it.

Quoted 4 lines

Ideally we'd unify things such that each architecture only needs *one* return trampoline that both ftrace and krpboes can use, which'd be significantly easier to get right and manage.

Agreed :-)

Thank you,

Quoted 3 lines

Thanks,
Mark.

Signature · 1 line
Masami Hiramatsu <[email protected]>
#5 lore raw

Re: [PATCH v13 bpf-next 1/1] rethook: x86: Add rethook x86 implementation

In reply to Peter Zijlstra
Quoted 10 lines · On Wed, 23 Mar 2022 13:34:54 +0100 Peter Zijlstra <[email protected]> wrote:
Quoted 5 lines · Masami Hiramatsu
Quoted 1 line

Also, what's rethook for anyway?

Rethook is a feature which hooks the function return. Most of the logic came from the kretprobe. Simply to say, 'kretprobe - kprobe' is the rethook :)

I got that far, but why did you take the bother to do these patches? Why wasn't 'use kretprobe' a good enough option?

Ah, sorry about lacking the background story.

Actually this came from Jiri's request of multiple kprobe for bpf[1]. He tried to solve an issue that starting bpf with multiple kprobe will take a long time because bpf-kprobe will wait for RCU grace period for sync rcu events.

Jiri wanted to attach a single bpf handler to multiple kprobes and he tried to introduce multiple-probe interface to kprobe. So I asked him to use ftrace and kretprobe-like hook if it is only for the function entry and exit, instead of adding ad-hoc interface
to kprobes. So I introduced fprobe (kprobe like interface for ftrace) and rethook (this is a generic return hook feature for fprobe exit handler)[2].

[1] https://lore.kernel.org/all/[email protected]/T/#u
[2] https://lore.kernel.org/all/164191321766.806991.7930388561276940676.stgit@devnote2/T/#u

This is the reason why I need to split the kretprobe's trampoline as rethook. Kretprobe is only for probing a single function entry/exit, thus it does not suit for this purpose.

Thank you,

Signature · 1 line
Masami Hiramatsu <[email protected]>
#12 lore raw
In reply to Masami Hiramatsu
Quoted 28 lines · Masami Hiramatsu
Quoted 11 lines · On Wed, 23 Mar 2022 14:18:40 +0000 Mark Rutland <[email protected]> wrote:
Quoted 1 line · Masami Hiramatsu

Hi,

Hi Masami,

Quoted 6 lines

Here is the 13th version of rethook x86 port. This is developed for a part of fprobe series [1] for hooking function return. But since I forgot to send it to arch maintainers, that caused conflict with IBT and SLS mitigation series. Now I picked the x86 rethook part and send it to x86 maintainers to be reviewed.

Quoted 5 lines
Quoted 1 line

[1] https://lore.kernel.org/all/164735281449.1084943.12438881786173547153.stgit@devnote2/T/#u

As mentioned elsewhere, I have similar (though not identical) concerns to Peter for the arm64 patch, which was equally unreviewed by maintainers, and the overall structure.

Yes, those should be reviewed by arch maintainers.

Quoted 6 lines
Quoted 5 lines

Note that this patch is still for the bpf-next since the rethook itself is on the bpf-next tree. But since this also uses the ANNOTATE_NOENDBR macro which has been introduced by IBT/ENDBR patch, to build this series you need to merge the tip/master branch with the bpf-next. (hopefully, it is rebased soon)

Quoted 7 lines
Quoted 4 lines

I thought we were going to drop the series from the bpf-next tree so that this could all go through review it had missed thusfar.

Is that still the plan? What's going on?

Now the arm64 (and other arch) port is reverted from bpf-next. I'll send those to you soon.

Ah; thanks for confirming!

Quoted 2 lines

Since bpf-next is focusing on x86 at first, I chose this for review in this version. Sorry for confusion.

No problem; I think the confusion is all my own, so nothing to apologise for! :)

Quoted 7 lines
Quoted 6 lines
Quoted 4 lines

The fprobe itself is for providing the function entry/exit probe with multiple probe point. The rethook is a sub-feature to hook the function return as same as kretprobe does. Eventually, I would like to replace the kretprobe's trampoline with this rethook.

Can we please start by converting each architecture to rethook?

Quoted 8 lines

Yes. As Peter pointed, I'm planning to add a kretprobe patches to use rethook if available in that series. let me prepare it.

Quoted 3 lines

Ideally we'd unify things such that each architecture only needs *one* return trampoline that both ftrace and krpboes can use, which'd be significantly easier to get right and manage.

Agreed :-)

Great!

Thanks,
Mark.

#6 lore raw

Re: [PATCH v13 bpf-next 1/1] rethook: x86: Add rethook x86 implementation

In reply to Masami Hiramatsu
Quoted 13 lines · Masami Hiramatsu
Quoted 6 lines · On Wed, 23 Mar 2022 09:05:26 +0100 Peter Zijlstra <[email protected]> wrote:
Quoted 2 lines · Masami Hiramatsu

Add rethook for x86 implementation. Most of the code has been copied from kretprobes on x86.

Right; as said, I'm really unhappy with growing a carbon copy of this stuff instead of sharing. Can we *please* keep it a single instance?

OK, then let me update the kprobe side too.

Quoted 7 lines
Quoted 2 lines

Them being basically indentical, it should be trivial to have CONFIG_KPROBE_ON_RETHOOK (or somesuch) and just share this.

Yes, ideally it should use CONFIG_HAVE_RETHOOK since the rethook arch port must be a copy of the kretprobe implementation. But for safety, I think having CONFIG_KPROBE_ON_RETHOOK is a good idea until replacing all kretprobe implementations.

Masami,

you're respinning this patch to combine arch_rethook_trampoline and __kretprobe_trampoline right?

#7 lore raw

Re: [PATCH v13 bpf-next 1/1] rethook: x86: Add rethook x86 implementation

Quoted 22 lines · On Thu, 24 Mar 2022 19:03:43 -0700 Alexei Starovoitov…
Quoted 13 lines · Masami Hiramatsu
Quoted 6 lines · On Wed, 23 Mar 2022 09:05:26 +0100 Peter Zijlstra <[email protected]> wrote:
Quoted 2 lines · Masami Hiramatsu

Add rethook for x86 implementation. Most of the code has been copied from kretprobes on x86.

Right; as said, I'm really unhappy with growing a carbon copy of this stuff instead of sharing. Can we *please* keep it a single instance?

OK, then let me update the kprobe side too.

Quoted 7 lines
Quoted 2 lines

Them being basically indentical, it should be trivial to have CONFIG_KPROBE_ON_RETHOOK (or somesuch) and just share this.

Yes, ideally it should use CONFIG_HAVE_RETHOOK since the rethook arch port must be a copy of the kretprobe implementation. But for safety, I think having CONFIG_KPROBE_ON_RETHOOK is a good idea until replacing all kretprobe implementations.

Quoted 5 lines

Masami,

you're respinning this patch to combine arch_rethook_trampoline and __kretprobe_trampoline right?

Yes, let me send the first patch set (for x86 at first).

BTW, can you review these 2 patches? These are only for the fprobes, so it can be picked to bpf-next.

https://lore.kernel.org/all/164802091567.1732982.1242854551611267542.stgit@devnote2/T/#u

Thank you,

Signature · 1 line
Masami Hiramatsu <[email protected]>
#8 lore raw

Re: [PATCH v13 bpf-next 1/1] rethook: x86: Add rethook x86 implementation

Quoted 26 lines · Masami Hiramatsu
Quoted 22 lines · On Thu, 24 Mar 2022 19:03:43 -0700 Alexei Starovoitov…
Quoted 20 lines · Masami Hiramatsu
Quoted 6 lines · On Wed, 23 Mar 2022 09:05:26 +0100 Peter Zijlstra <[email protected]> wrote:
Quoted 2 lines · Masami Hiramatsu

Add rethook for x86 implementation. Most of the code has been copied from kretprobes on x86.

Right; as said, I'm really unhappy with growing a carbon copy of this stuff instead of sharing. Can we *please* keep it a single instance?

OK, then let me update the kprobe side too.

Quoted 2 lines

Them being basically indentical, it should be trivial to have CONFIG_KPROBE_ON_RETHOOK (or somesuch) and just share this.

Yes, ideally it should use CONFIG_HAVE_RETHOOK since the rethook arch port must be a copy of the kretprobe implementation. But for safety, I think having CONFIG_KPROBE_ON_RETHOOK is a good idea until replacing all kretprobe implementations.

Quoted 7 lines
Quoted 5 lines

Masami,

you're respinning this patch to combine arch_rethook_trampoline and __kretprobe_trampoline right?

Yes, let me send the first patch set (for x86 at first).

great

Quoted 4 lines

BTW, can you review these 2 patches? These are only for the fprobes, so it can be picked to bpf-next.

https://lore.kernel.org/all/164802091567.1732982.1242854551611267542.stgit@devnote2/T/#u

Yes. They look good. Will push them soon.