Skip to contentSkip to conversation

arm64: kprobes: Use BRK instead of single-step when executing instructions out-of-line

PATCH Applied Jean-Philippe Brucker 3 participants 5 messages Nov 3, 2020 – Nov 10, 2020 arm64

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/2 arm64: kprobes: Use BRK instead of single-step when executing instructions out-of-line

Commit 36dadef23fcc ("kprobes: Init kprobes in early_initcall") enabled using kprobes from early_initcall. Unfortunately at this point the hardware debug infrastructure is not operational. The OS lock may still be locked, and the hardware watchpoints may have unknown values when kprobe enables debug monitors to single-step instructions.

Rather than using hardware single-step, append a BRK instruction after the instruction to be executed out-of-line.

Fixes 36dadef23fcc ("kprobes: Init kprobes in early_initcall")

---
v2:
* Remove patch_text() helper
* Mask all of DAIF during trampoline
v1: https://lore.kernel.org/linux-arm-kernel/[email protected]/

---
 arch/arm64/include/asm/brk-imm.h        |  2 +
 arch/arm64/include/asm/debug-monitors.h |  1 +
 arch/arm64/include/asm/kprobes.h        |  2 +-
 arch/arm64/kernel/probes/kprobes.c      | 69 +++++++++----------------
 4 files changed, 27 insertions(+), 47 deletions(-)
arch/arm64/include/asm/brk-imm.h +2 -0
index e3d47b52161d..ec7720dbe2c8 100644
@@ -10,6 +10,7 @@
  * #imm16 values used for BRK instruction generation
  * 0x004: for installing kprobes
  * 0x005: for installing uprobes
+ * 0x006: for kprobe software single-step
  * Allowed values for kgdb are 0x400 - 0x7ff
  * 0x100: for triggering a fault on purpose (reserved)
  * 0x400: for dynamic BRK instruction
@@ -19,6 +20,7 @@
  */
 #define KPROBES_BRK_IMM			0x004
 #define UPROBES_BRK_IMM			0x005
+#define KPROBES_BRK_SS_IMM		0x006
 #define FAULT_BRK_IMM			0x100
 #define KGDB_DYN_DBG_BRK_IMM		0x400
 #define KGDB_COMPILED_DBG_BRK_IMM	0x401
arch/arm64/include/asm/debug-monitors.h +1 -0
index 0b298f48f5bf..657c921fd784 100644
@@ -53,6 +53,7 @@
 
 /* kprobes BRK opcodes with ESR encoding  */
 #define BRK64_OPCODE_KPROBES	(AARCH64_BREAK_MON | (KPROBES_BRK_IMM << 5))
+#define BRK64_OPCODE_KPROBES_SS	(AARCH64_BREAK_MON | (KPROBES_BRK_SS_IMM << 5))
 /* uprobes BRK opcodes with ESR encoding  */
 #define BRK64_OPCODE_UPROBES	(AARCH64_BREAK_MON | (UPROBES_BRK_IMM << 5))
 
arch/arm64/include/asm/kprobes.h +1 -1
index 97e511d645a2..8699ce30f587 100644
@@ -16,7 +16,7 @@
 #include <linux/percpu.h>
 
 #define __ARCH_WANT_KPROBES_INSN_SLOT
-#define MAX_INSN_SIZE			1
+#define MAX_INSN_SIZE			2
 
 #define flush_insn_slot(p)		do { } while (0)
 #define kretprobe_blacklist_size	0
arch/arm64/kernel/probes/kprobes.c +23 -46

151 lines not on this page · raw message

Signature · 7 lines
2.29.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#1 lore raw

PATCH v2 2/2 arm64: kprobes: Remove redundant kprobe_step_ctx

From: Masami Hiramatsu <[email protected]>

The kprobe_step_ctx (kcb->ss_ctx) has ss_pending and match_addr, but those are redundant because those can be replaced by KPROBE_HIT_SS and &cur_kprobe->ainsn.api.insn[1] respectively.
To simplify the code, remove the kprobe_step_ctx.

---
 arch/arm64/include/asm/kprobes.h   |  7 ----
 arch/arm64/kernel/probes/kprobes.c | 53 +++++++-----------------------
 2 files changed, 12 insertions(+), 48 deletions(-)
arch/arm64/include/asm/kprobes.h +0 -7
index 8699ce30f587..5d38ff4a4806 100644
@@ -28,18 +28,11 @@ struct prev_kprobe {
 	unsigned int status;
 };
 
-/* Single step context for kprobe */
-struct kprobe_step_ctx {
-	unsigned long ss_pending;
-	unsigned long match_addr;
-};
-
 /* per-cpu kprobe control block */
 struct kprobe_ctlblk {
 	unsigned int kprobe_status;
 	unsigned long saved_irqflag;
 	struct prev_kprobe prev_kprobe;
-	struct kprobe_step_ctx ss_ctx;
 };
 
 void arch_remove_kprobe(struct kprobe *);
arch/arm64/kernel/probes/kprobes.c +12 -41

98 lines not on this page · raw message

Signature · 7 lines
2.29.1


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
maintainer Acked-by on v2 2/2
#2 lore raw

Re: [PATCH v2 2/2] arm64: kprobes: Remove redundant kprobe_step_ctx

Quoted 7 lines · Jean-Philippe Brucker

From: Masami Hiramatsu <[email protected]>

The kprobe_step_ctx (kcb->ss_ctx) has ss_pending and match_addr, but those are redundant because those can be replaced by KPROBE_HIT_SS and &cur_kprobe->ainsn.api.insn[1] respectively.
To simplify the code, remove the kprobe_step_ctx.

Quoted 6 lines
---
 arch/arm64/include/asm/kprobes.h   |  7 ----
 arch/arm64/kernel/probes/kprobes.c | 53 +++++++-----------------------
 2 files changed, 12 insertions(+), 48 deletions(-)

Catalin should be able to queue this at -rc3, once the first patch has landed in mainline.

Cheers,

Will

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

maintainer Applied
#3 lore raw

Re: [PATCH v2 2/2] arm64: kprobes: Remove redundant kprobe_step_ctx

Quoted 15 lines · Will Deacon
Quoted 7 lines · Jean-Philippe Brucker

From: Masami Hiramatsu <[email protected]>

The kprobe_step_ctx (kcb->ss_ctx) has ss_pending and match_addr, but those are redundant because those can be replaced by KPROBE_HIT_SS and &cur_kprobe->ainsn.api.insn[1] respectively.
To simplify the code, remove the kprobe_step_ctx.

Quoted 6 lines
---
 arch/arm64/include/asm/kprobes.h   |  7 ----
 arch/arm64/kernel/probes/kprobes.c | 53 +++++++-----------------------
 2 files changed, 12 insertions(+), 48 deletions(-)
Quoted 4 lines

Catalin should be able to queue this at -rc3, once the first patch has landed in mainline.

I queued it but couldn't convince "b4 ty" to reply to patch 2:

https://git.kernel.org/arm64/c/ba090f9cafd53dbabe0f0a8c4ccae44203d3731b

Signature · 6 lines
Catalin

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
maintainer
#4 lore raw
In reply to Jean-Philippe Brucker
Quoted 8 lines · Jean-Philippe Brucker

Commit 36dadef23fcc ("kprobes: Init kprobes in early_initcall") enabled using kprobes from early_initcall. Unfortunately at this point the hardware debug infrastructure is not operational. The OS lock may still be locked, and the hardware watchpoints may have unknown values when kprobe enables debug monitors to single-step instructions.

Rather than using hardware single-step, append a BRK instruction after the instruction to be executed out-of-line.

Applied first patch to arm64 (for-next/fixes), thanks!

[1/2] arm64: kprobes: Use BRK instead of single-step when executing instructions out-of-line

      https://git.kernel.org/arm64/c/7ee31a3aa8f4

Cheers,

Signature · 10 lines
Will

https://fixes.arm64.dev
https://next.arm64.dev
https://will.arm64.dev

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel