Kprobes: wrapper to define jprobe.entry
PATCH 1/1 Kprobes: wrapper to define jprobe.entry
Hi,
Here is a patch that adds a wrapper for defining jprobe.entry to make it easy to handle the three dword function descriptors defined by the PowerPC ELF ABI.
Current patch against 2.6.10-rc2-mm1 + kprobes patch for ppc64.
Changes for adding this wrapper for x86, ppc64 (tested) and x86_64 (untested) below. The earlier method of defining jprobe.entry will continue to work.
Here is a pseudocode snippet to use jprobes with this patch.
............
struct jprobe jp;
jtcp_v4_rcv(struct skbuff *skb)
{
/* decode and log skb related details as required */
jprobe_return();
return 0;
}init_module
{
jp.kp.addr = (kprobe_opcode_t *)<addr of tcp_v4_rcv>;
jp.entry = JPROBE_ENTRY(jtcp_v4_rcv);
register_jprobe(&jp);
return 0;
}cleanup_module
{
unregister_jprobe(&jp);
}............
Dave,
I am not aware of the semantics for sparc64 for making this change.
Thanks,
Ananth
include/asm-i386/kprobes.h renamed from linux-2.6.10-rc2/include/asm-i386/kprobes.h +2 -0
@@ -38,6 +38,8 @@ typedef u8 kprobe_opcode_t; ? (MAX_STACK_SIZE) \ : (((unsigned long)current_thread_info()) + THREAD_SIZE - (ADDR))) +#define JPROBE_ENTRY (pentry) (kprobe_opcode_t *)pentry +/* Architecture specific copy of original instruction*/ struct arch_specific_insn {/* copy of the original instruction */
include/asm-ppc64/kprobes.h renamed from linux-2.6.10-rc2/include/asm-ppc64/kprobes.h +2 -0
@@ -35,6 +35,8 @@ typedef unsigned int kprobe_opcode_t;#define BREAKPOINT_INSTRUCTION 0x7fe00008/* trap */ #define MAX_INSN_SIZE 1 +#define JPROBE_ENTRY (pentry) (kprobe_opcode_t *)((func_descr_t *)pentry) +/* Architecture specific copy of original instruction */ struct arch_specific_insn {/* copy of original instruction */
include/asm-x86_64/kprobes.h renamed from linux-2.6.10-rc2/include/asm-x86_64/kprobes.h +2 -0
@@ -37,6 +37,8 @@ typedef u8 kprobe_opcode_t; ? (MAX_STACK_SIZE) \ : (((unsigned long)current_thread_info()) + THREAD_SIZE - (ADDR))) +#define JPROBE_ENTRY (pentry) (kprobe_opcode_t *)pentry +/* Architecture specific copy of original instruction*/ struct arch_specific_insn {/* copy of the original instruction */
Quoted 6 lines · Ananth N Mavinakayanahalli
Here is a patch that adds a wrapper for defining jprobe.entry to make it easy to handle the three dword function descriptors defined by the PowerPC ELF ABI.
Current patch against 2.6.10-rc2-mm1 + kprobes patch for ppc64.
I don't have the kprobes-for-ppc64 patch here.
Quoted 3 lines
Changes for adding this wrapper for x86, ppc64 (tested) and
x86_64(untested) below. The earlier method of defining jprobe.entry will continue to work.
So what should I do with this? I'm inclined to drop it until the x86_64 part has been tested and Dave has had a go at the sparc64 version.
Quoted 7 lines · On Thu, 18 Nov 2004 14:47:46 -0800 Andrew Morton <[email protected]> wrote:
Quoted 3 lines · Ananth N Mavinakayanahalli
Changes for adding this wrapper for x86, ppc64 (tested) and
x86_64(untested) below. The earlier method of defining jprobe.entry will continue to work.So what should I do with this? I'm inclined to drop it until the
x86_64part has been tested and Dave has had a go at the sparc64 version.
Yes, now that we have kprobe support on 4 platforms, it is important that anyone who changes public parts of this interface do the necessary per-platform fixups necessary to coincide with such changes.
I think the person changing the data type should be the one fixing up sparc64 :-)
Quoted 1 line · Andrew Morton
Ananth N Mavinakayanahalli <[email protected]> wrote:
Hi Andrew,
Quoted 7 lines
Quoted 6 lines
Here is a patch that adds a wrapper for defining jprobe.entry to make it easy to handle the three dword function descriptors defined by the PowerPC ELF ABI.
Current patch against 2.6.10-rc2-mm1 + kprobes patch for ppc64.
Quoted 8 lines
I don't have the kprobes-for-ppc64 patch here.
Quoted 3 lines
Changes for adding this wrapper for x86, ppc64 (tested) and
x86_64(untested) below. The earlier method of defining jprobe.entry will continue to work.So what should I do with this? I'm inclined to drop it until the
x86_64part has been tested and Dave has had a go at the sparc64 version.
I have now tested the patch succesfully on x86_64 and updated it for sparc64 too (Dave says the change looks good).
Please apply.
Thanks,
Ananth
include/asm-i386/kprobes.h renamed from linux-2.6.10-rc2/include/asm-i386/kprobes.h +2 -0
@@ -38,6 +38,8 @@ typedef u8 kprobe_opcode_t; ? (MAX_STACK_SIZE) \ : (((unsigned long)current_thread_info()) + THREAD_SIZE - (ADDR))) +#define JPROBE_ENTRY (pentry) (kprobe_opcode_t *)pentry +/* Architecture specific copy of original instruction*/ struct arch_specific_insn {/* copy of the original instruction */
include/asm-ppc64/kprobes.h renamed from linux-2.6.10-rc2/include/asm-ppc64/kprobes.h +2 -0
@@ -35,6 +35,8 @@ typedef unsigned int kprobe_opcode_t;#define BREAKPOINT_INSTRUCTION 0x7fe00008/* trap */ #define MAX_INSN_SIZE 1 +#define JPROBE_ENTRY (pentry) (kprobe_opcode_t *)((func_descr_t *)pentry) +/* Architecture specific copy of original instruction */ struct arch_specific_insn {/* copy of original instruction */
include/asm-sparc64/kprobes.h renamed from linux-2.6.10-rc2/include/asm-sparc64/kprobes.h +2 -0
@@ -10,6 +10,8 @@ typedef u32 kprobe_opcode_t;#define BREAKPOINT_INSTRUCTION_2 0x91d02071/* ta 0x71 */ #define MAX_INSN_SIZE 2 +#define JPROBE_ENTRY (pentry) (kprobe_opcode_t *)pentry +/* Architecture specific copy of original instruction*/ struct arch_specific_insn {/* copy of the original instruction */
include/asm-x86_64/kprobes.h renamed from linux-2.6.10-rc2/include/asm-x86_64/kprobes.h +2 -0
@@ -37,6 +37,8 @@ typedef u8 kprobe_opcode_t; ? (MAX_STACK_SIZE) \ : (((unsigned long)current_thread_info()) + THREAD_SIZE - (ADDR))) +#define JPROBE_ENTRY (pentry) (kprobe_opcode_t *)pentry +/* Architecture specific copy of original instruction*/ struct arch_specific_insn {/* copy of the original instruction */
Quoted 14 lines · Ananth N Mavinakayanahalli
Quoted 14 lines
Quoted 7 lines
Here is a patch that adds a wrapper for defining jprobe.entry to make it easy to handle the three dword function descriptors defined by the PowerPC ELF ABI.
Current patch against 2.6.10-rc2-mm1 + kprobes patch for ppc64.
I don't have the kprobes-for-ppc64 patch here.
Quoted 3 lines
Changes for adding this wrapper for x86, ppc64 (tested) and
x86_64(untested) below. The earlier method of defining jprobe.entry will continue to work.
Quoted 5 lines
Quoted 2 lines
So what should I do with this? I'm inclined to drop it until the
x86_64part has been tested and Dave has had a go at the sparc64 version.I have now tested the patch succesfully on
x86_64and updated it for sparc64 too (Dave says the change looks good).
What is the review and testing status of the kprobes-for-ppc64 patch which you sent?
Quoted 15 lines · Andrew Morton
Quoted 14 lines · Ananth N Mavinakayanahalli
Quoted 14 lines
Quoted 7 lines
Here is a patch that adds a wrapper for defining jprobe.entry to make it easy to handle the three dword function descriptors defined by the PowerPC ELF ABI.
Current patch against 2.6.10-rc2-mm1 + kprobes patch for ppc64.
I don't have the kprobes-for-ppc64 patch here.
Quoted 3 lines
Changes for adding this wrapper for x86, ppc64 (tested) and
x86_64(untested) below. The earlier method of defining jprobe.entry will continue to work.
Quoted 9 lines
Quoted 5 lines
Quoted 2 lines
So what should I do with this? I'm inclined to drop it until the
x86_64part has been tested and Dave has had a go at the sparc64 version.I have now tested the patch succesfully on
x86_64and updated it for sparc64 too (Dave says the change looks good).What is the review and testing status of the kprobes-for-ppc64 patch which you sent?
The patch was earlier posted on the PPC64 mailing list for comments and Paul had reviewed it. I had to update the patch to take care of the base kprobe changes that were made to accomodate the x86_64 port.
The patch is tested on POWER3 (uni) and POWER4 (lpar).
Thanks,
Ananth