FAILED: patch "[PATCH] tracing/uprobes: Fix output for multiple string arguments" failed to apply to 4.19-stable tree
Subsystems (2)
[email protected]
maintainerThe patch below does not apply to the 4.19-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.
thanks,
greg k-h
------------------ original commit in Linus's tree ------------------
From 0722069a5374b904ec1a67f91249f90e1cfae259 Mon Sep 17 00:00:00 2001
From: Andreas Ziegler <[email protected]>
Date: Wed, 16 Jan 2019 15:16:29 +0100
Subject: [PATCH] tracing/uprobes: Fix output for multiple string arguments
When printing multiple uprobe arguments as strings the output for the earlier arguments would also include all later string arguments.
This is best explained in an example:
Consider adding a uprobe to a function receiving two strings as parameters which is at offset 0xa0 in strlib.so and we want to print both parameters when the uprobe is hit (on x86_64):
$ echo 'p:func /lib/strlib.so:0xa0 +0(%di):string +0(%si):string' > \
/sys/kernel/debug/tracing/uprobe_eventsWhen the function is called as func("foo", "bar") and we hit the probe, the trace file shows a line like the following:
[...] func: (0x7f7e683706a0) arg1="foobar" arg2="bar"
Note the extra "bar" printed as part of arg1. This behaviour stacks up for additional string arguments.
The strings are stored in a dynamically growing part of the uprobe buffer by fetch_store_string() after copying them from userspace via strncpy_from_user(). The return value of strncpy_from_user() is then directly used as the required size for the string. However, this does not take the terminating null byte into account as the documentation for strncpy_from_user() cleary states that it "[...] returns the length of the string (not including the trailing NUL)" even though the null byte will be copied to the destination.
Therefore, subsequent calls to fetch_store_string() will overwrite the terminating null byte of the most recently fetched string with the first character of the current string, leading to the "accumulation" of strings in earlier arguments in the output.
Fix this by incrementing the return value of strncpy_from_user() by one if we did not hit the maximum buffer size.
kernel/trace/trace_uprobe.c +7 -0
@@ -160,6 +160,13 @@ fetch_store_string(unsigned long addr, void *dest, void *base)if (ret >=0 ) {if (ret == maxlen) dst[ret -1 ] ='\0' ; +else +/* +* Include the terminating null byte. In this case it +* was copied by strncpy_from_user but not accounted +* for in ret. +*/ + ret++; *(u32 *)dest =make_data_loc (ret, (void *)dst - base); }
Masami Hiramatsu
maintainerPATCH 1/1 tracing/uprobes: Fix output for multiple string arguments
From: Andreas Ziegler <[email protected]>
commit 0722069a5374b904ec1a67f91249f90e1cfae259 upstream.
When printing multiple uprobe arguments as strings the output for the earlier arguments would also include all later string arguments.
This is best explained in an example:
Consider adding a uprobe to a function receiving two strings as parameters which is at offset 0xa0 in strlib.so and we want to print both parameters when the uprobe is hit (on x86_64):
$ echo 'p:func /lib/strlib.so:0xa0 +0(%di):string +0(%si):string' > \
/sys/kernel/debug/tracing/uprobe_eventsWhen the function is called as func("foo", "bar") and we hit the probe, the trace file shows a line like the following:
[...] func: (0x7f7e683706a0) arg1="foobar" arg2="bar"
Note the extra "bar" printed as part of arg1. This behaviour stacks up for additional string arguments.
The strings are stored in a dynamically growing part of the uprobe buffer by fetch_store_string() after copying them from userspace via strncpy_from_user(). The return value of strncpy_from_user() is then directly used as the required size for the string. However, this does not take the terminating null byte into account as the documentation for strncpy_from_user() cleary states that it "[...] returns the length of the string (not including the trailing NUL)" even though the null byte will be copied to the destination.
Therefore, subsequent calls to fetch_store_string() will overwrite the terminating null byte of the most recently fetched string with the first character of the current string, leading to the "accumulation" of strings in earlier arguments in the output.
Fix this by incrementing the return value of strncpy_from_user() by one if we did not hit the maximum buffer size.
--- kernel/trace/trace_uprobe.c | 7 +++++++ 1 file changed, 7 insertions(+)
kernel/trace/trace_uprobe.c +7 -0
@@ -142,6 +142,13 @@ static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, ret =strncpy_from_user (dst, src, maxlen);if (ret == maxlen) dst[--ret] ='\0' ; +else if (ret >=0 ) +/* +* Include the terminating null byte. In this case it +* was copied by strncpy_from_user but not accounted +* for in ret. +*/ + ret++;if (ret <0 ) {/* Failed to fetch string */ ((u8 *)get_rloc_data (dest))[0 ] ='\0' ;
Re: [PATCH stable-4.19] tracing/uprobes: Fix output for multiple string arguments
Hi,
kernel/trace/trace_uprobe.c static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, · line 145
Quoted 10 lines
kernel/trace/trace_uprobe.c +1
-0@@ -142,6 +142,13 @@ static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, ret =strncpy_from_user (dst, src, maxlen);if (ret == maxlen) dst[--ret] ='\0' ; +else if (ret >=0 )
with this fix the maxlen case still decrements ret by one which does not include the terminating \0 (and leaves maxlen = 1 in store_trace_args()). Usually this only happens with the last element but I still feel that it's inconsistent...
Would'nt it be easier to simply add a '+ 1' to to the make_rloc_data() statement, like this:
kernel/trace/trace_uprobe.c +1 -1
@@ -147,7 +147,7 @@ static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, ((u8 *)get_rloc_data (dest))[0 ] ='\0' ; *(u32 *)dest =make_data_rloc (0 ,get_rloc_offs (rloc)); }else { - *(u32 *)dest =make_data_rloc (ret,get_rloc_offs (rloc)); + *(u32 *)dest =make_data_rloc (ret +1 ,get_rloc_offs (rloc)); } }
kernel/trace/trace_uprobe.c static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, · line 154
Quoted 10 lines
kernel/trace/trace_uprobe.c +6
-0+/* +* Include the terminating null byte. In this case it +* was copied by strncpy_from_user but not accounted +* for in ret. +*/ + ret++;if (ret <0 ) {/* Failed to fetch string */ ((u8 *)get_rloc_data (dest))[0 ] ='\0' ;
Regards,
Andreas
Masami Hiramatsu
maintainerRe: [PATCH stable-4.19] tracing/uprobes: Fix output for multiple string arguments
Hi Andreas,
Quoted 17 lines · On Thu, 14 Feb 2019 14:00:03 +0100 Andreas Ziegler <[email protected]>…
Hi,
kernel/trace/trace_uprobe.c static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, · line 145
Quoted 10 lines
kernel/trace/trace_uprobe.c +1
-0@@ -142,6 +142,13 @@ static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, ret =strncpy_from_user (dst, src, maxlen);if (ret == maxlen) dst[--ret] ='\0' ; +else if (ret >=0 )with this fix the maxlen case still decrements ret by one which does not include the terminating \0 (and leaves maxlen = 1 in
store_trace_args()). Usually this only happens with the last element but I still feel that it's inconsistent...
Yes, I found that too. And I thought it was another fix.
Quoted 3 lines
Would'nt it be easier to simply add a '+ 1' to to the
make_rloc_data()statement, like this:
Hmm, in that case, I would like to make "--ret" to "ret - 1" as the latest kernel does. Keeping the final code similar will help us if we need another fix to backport.
Any thought?
Thank you,
Quoted 29 lines
kernel/trace/trace_uprobe.c +1
-1@@ -147,7 +147,7 @@ static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, ((u8 *)get_rloc_data (dest))[0 ] ='\0' ; *(u32 *)dest =make_data_rloc (0 ,get_rloc_offs (rloc)); }else { - *(u32 *)dest =make_data_rloc (ret,get_rloc_offs (rloc)); + *(u32 *)dest =make_data_rloc (ret +1 ,get_rloc_offs (rloc)); } }kernel/trace/trace_uprobe.c static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, · line 154
Quoted 11 lines
kernel/trace/trace_uprobe.c +6
-0+/* +* Include the terminating null byte. In this case it +* was copied by strncpy_from_user but not accounted +* for in ret. +*/ + ret++;if (ret <0 ) {/* Failed to fetch string */ ((u8 *)get_rloc_data (dest))[0 ] ='\0' ;Regards,
Andreas
Signature · 1 line
Masami Hiramatsu <[email protected]>
Masami Hiramatsu
maintainerPATCH v2 1/1 tracing/uprobes: Fix output for multiple string arguments
From: Andreas Ziegler <[email protected]>
commit 0722069a5374b904ec1a67f91249f90e1cfae259 upstream.
When printing multiple uprobe arguments as strings the output for the earlier arguments would also include all later string arguments.
This is best explained in an example:
Consider adding a uprobe to a function receiving two strings as parameters which is at offset 0xa0 in strlib.so and we want to print both parameters when the uprobe is hit (on x86_64):
$ echo 'p:func /lib/strlib.so:0xa0 +0(%di):string +0(%si):string' > \
/sys/kernel/debug/tracing/uprobe_eventsWhen the function is called as func("foo", "bar") and we hit the probe, the trace file shows a line like the following:
[...] func: (0x7f7e683706a0) arg1="foobar" arg2="bar"
Note the extra "bar" printed as part of arg1. This behaviour stacks up for additional string arguments.
The strings are stored in a dynamically growing part of the uprobe buffer by fetch_store_string() after copying them from userspace via strncpy_from_user(). The return value of strncpy_from_user() is then directly used as the required size for the string. However, this does not take the terminating null byte into account as the documentation for strncpy_from_user() cleary states that it "[...] returns the length of the string (not including the trailing NUL)" even though the null byte will be copied to the destination.
Therefore, subsequent calls to fetch_store_string() will overwrite the terminating null byte of the most recently fetched string with the first character of the current string, leading to the "accumulation" of strings in earlier arguments in the output.
Fix this by incrementing the return value of strncpy_from_user() by one if we did not hit the maximum buffer size.
--- kernel/trace/trace_uprobe.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)
kernel/trace/trace_uprobe.c +8 -1
@@ -141,7 +141,14 @@ static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, ret =strncpy_from_user (dst, src, maxlen);if (ret == maxlen) - dst[--ret] ='\0' ; + dst[ret -1 ] ='\0' ; +else if (ret >=0 ) +/* +* Include the terminating null byte. In this case it +* was copied by strncpy_from_user but not accounted +* for in ret. +*/ + ret++;if (ret <0 ) {/* Failed to fetch string */ ((u8 *)get_rloc_data (dest))[0 ] ='\0' ;
Masami Hiramatsu
maintainerRe: [PATCH stable-4.19 v2] tracing/uprobes: Fix output for multiple string arguments
Hi Andreas,
I've update this patch to include "ret == maxlen" case fix as similar to the latest kernel code.
Please review it. If it is OK, I'll update patches for other stable trees.
Thank you,
Quoted 75 lines · On Thu, 14 Feb 2019 23:24:29 +0900 Masami Hiramatsu <[email protected]> wrote:
From: Andreas Ziegler <[email protected]>
commit
0722069a5374b904ec1a67f91249f90e1cfae259upstream.When printing multiple uprobe arguments as strings the output for the earlier arguments would also include all later string arguments.
This is best explained in an example:
Consider adding a uprobe to a function receiving two strings as parameters which is at offset
0xa0in strlib.so and we want to print both parameters when the uprobe is hit (onx86_64):$ echo 'p:func /lib/strlib.so:0xa0 +0(%di):string +0(%si):string' > \ /sys/kernel/debug/tracing/uprobe_eventsWhen the function is called as
func("foo", "bar")and we hit the probe, the trace file shows a line like the following:[...] func: (
0x7f7e683706a0) arg1="foobar" arg2="bar"Note the extra "bar" printed as part of arg1. This behaviour stacks up for additional string arguments.
The strings are stored in a dynamically growing part of the uprobe buffer by
fetch_store_string()after copying them from userspace viastrncpy_from_user(). The return value ofstrncpy_from_user()is then directly used as the required size for the string. However, this does not take the terminating null byte into account as the documentation forstrncpy_from_user()cleary states that it "[...] returns the length of the string (not including the trailing NUL)" even though the null byte will be copied to the destination.Therefore, subsequent calls to
fetch_store_string()will overwrite the terminating null byte of the most recently fetched string with the first character of the current string, leading to the "accumulation" of strings in earlier arguments in the output.Fix this by incrementing the return value of
strncpy_from_user()by one if we did not hit the maximum buffer size.Cc Ingo Molnar <[email protected]>Fixes 5baaa59ef09e ("tracing/probes: Implement 'memory' fetch method for uprobes")Acked-by Masami Hiramatsu <[email protected]>Signed-off-by Andreas Ziegler <[email protected]>Signed-off-by Steven Rostedt (VMware) <[email protected]>Signed-off-by Masami Hiramatsu <[email protected]>--- kernel/trace/trace_uprobe.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)kernel/trace/trace_uprobe.c +8
-1@@ -141,7 +141,14 @@ static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, ret =strncpy_from_user (dst, src, maxlen);if (ret == maxlen) - dst[--ret] ='\0' ; + dst[ret -1 ] ='\0' ; +else if (ret >=0 ) +/* +* Include the terminating null byte. In this case it +* was copied by strncpy_from_user but not accounted +* for in ret. +*/ + ret++;if (ret <0 ) {/* Failed to fetch string */ ((u8 *)get_rloc_data (dest))[0 ] ='\0' ;
Signature · 1 line
Masami Hiramatsu <[email protected]>
Re: [PATCH stable-4.19 v2] tracing/uprobes: Fix output for multiple string arguments
Hi Masami,
yes, that patch looks like a good compromise between the original code and the new patch. To reduce the number of if statements involving ret we would need to move more things around and I think it's best to avoid that (as you also said).
I'd say you can update this for the other kernel stable trees as well.
Thank you!
Andreas
P.S.: Do we need to explain in the commit message why backporting is needed?
Quoted 94 lines · Masami Hiramatsu
Hi Andreas,
I've update this patch to include "ret == maxlen" case fix as similar to the latest kernel code.
Please review it. If it is OK, I'll update patches for other stable trees.
Thank you,
Quoted 75 lines · On Thu, 14 Feb 2019 23:24:29 +0900 Masami Hiramatsu <[email protected]> wrote:
From: Andreas Ziegler <[email protected]>
commit
0722069a5374b904ec1a67f91249f90e1cfae259upstream.When printing multiple uprobe arguments as strings the output for the earlier arguments would also include all later string arguments.
This is best explained in an example:
Consider adding a uprobe to a function receiving two strings as parameters which is at offset
0xa0in strlib.so and we want to print both parameters when the uprobe is hit (onx86_64):$ echo 'p:func /lib/strlib.so:0xa0 +0(%di):string +0(%si):string' > \ /sys/kernel/debug/tracing/uprobe_eventsWhen the function is called as
func("foo", "bar")and we hit the probe, the trace file shows a line like the following:[...] func: (
0x7f7e683706a0) arg1="foobar" arg2="bar"Note the extra "bar" printed as part of arg1. This behaviour stacks up for additional string arguments.
The strings are stored in a dynamically growing part of the uprobe buffer by
fetch_store_string()after copying them from userspace viastrncpy_from_user(). The return value ofstrncpy_from_user()is then directly used as the required size for the string. However, this does not take the terminating null byte into account as the documentation forstrncpy_from_user()cleary states that it "[...] returns the length of the string (not including the trailing NUL)" even though the null byte will be copied to the destination.Therefore, subsequent calls to
fetch_store_string()will overwrite the terminating null byte of the most recently fetched string with the first character of the current string, leading to the "accumulation" of strings in earlier arguments in the output.Fix this by incrementing the return value of
strncpy_from_user()by one if we did not hit the maximum buffer size.Cc Ingo Molnar <[email protected]>Fixes 5baaa59ef09e ("tracing/probes: Implement 'memory' fetch method for uprobes")Acked-by Masami Hiramatsu <[email protected]>Signed-off-by Andreas Ziegler <[email protected]>Signed-off-by Steven Rostedt (VMware) <[email protected]>Signed-off-by Masami Hiramatsu <[email protected]>---
kernel/trace/trace_uprobe.c| 9 ++++++++-1 file changed, 8 insertions(+), 1 deletion(-)kernel/trace/trace_uprobe.c +8
-1@@ -141,7 +141,14 @@ static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, ret =strncpy_from_user (dst, src, maxlen);if (ret == maxlen) - dst[--ret] ='\0' ; + dst[ret -1 ] ='\0' ; +else if (ret >=0 ) +/* +* Include the terminating null byte. In this case it +* was copied by strncpy_from_user but not accounted +* for in ret. +*/ + ret++;if (ret <0 ) {/* Failed to fetch string */ ((u8 *)get_rloc_data (dest))[0 ] ='\0' ;Signature · 1 line
Masami Hiramatsu <[email protected]>
Masami Hiramatsu
maintainerRe: [PATCH stable-4.19 v2] tracing/uprobes: Fix output for multiple string arguments
Hi Andreas,
Quoted 9 lines · On Thu, 14 Feb 2019 16:53:26 +0100 Andreas Ziegler <[email protected]>…
Hi Masami,
yes, that patch looks like a good compromise between the original code and the new patch. To reduce the number of if statements involving ret we would need to move more things around and I think it's best to avoid that (as you also said).
I'd say you can update this for the other kernel stable trees as well.
Thank you very much!
Quoted 103 lines
Thank you!
Andreas
P.S.: Do we need to explain in the commit message why backporting is needed?Quoted 94 lines · Masami Hiramatsu
Hi Andreas,
I've update this patch to include "ret == maxlen" case fix as similar to the latest kernel code.
Please review it. If it is OK, I'll update patches for other stable trees.
Thank you,
Quoted 75 lines · On Thu, 14 Feb 2019 23:24:29 +0900 Masami Hiramatsu <[email protected]> wrote:
From: Andreas Ziegler <[email protected]>
commit
0722069a5374b904ec1a67f91249f90e1cfae259upstream.When printing multiple uprobe arguments as strings the output for the earlier arguments would also include all later string arguments.
This is best explained in an example:
Consider adding a uprobe to a function receiving two strings as parameters which is at offset
0xa0in strlib.so and we want to print both parameters when the uprobe is hit (onx86_64):$ echo 'p:func /lib/strlib.so:0xa0 +0(%di):string +0(%si):string' > \ /sys/kernel/debug/tracing/uprobe_eventsWhen the function is called as
func("foo", "bar")and we hit the probe, the trace file shows a line like the following:[...] func: (
0x7f7e683706a0) arg1="foobar" arg2="bar"Note the extra "bar" printed as part of arg1. This behaviour stacks up for additional string arguments.
The strings are stored in a dynamically growing part of the uprobe buffer by
fetch_store_string()after copying them from userspace viastrncpy_from_user(). The return value ofstrncpy_from_user()is then directly used as the required size for the string. However, this does not take the terminating null byte into account as the documentation forstrncpy_from_user()cleary states that it "[...] returns the length of the string (not including the trailing NUL)" even though the null byte will be copied to the destination.Therefore, subsequent calls to
fetch_store_string()will overwrite the terminating null byte of the most recently fetched string with the first character of the current string, leading to the "accumulation" of strings in earlier arguments in the output.Fix this by incrementing the return value of
strncpy_from_user()by one if we did not hit the maximum buffer size.Cc Ingo Molnar <[email protected]>Fixes 5baaa59ef09e ("tracing/probes: Implement 'memory' fetch method for uprobes")Acked-by Masami Hiramatsu <[email protected]>Signed-off-by Andreas Ziegler <[email protected]>Signed-off-by Steven Rostedt (VMware) <[email protected]>Signed-off-by Masami Hiramatsu <[email protected]>---
kernel/trace/trace_uprobe.c| 9 ++++++++-1 file changed, 8 insertions(+), 1 deletion(-)kernel/trace/trace_uprobe.c +8
-1@@ -141,7 +141,14 @@ static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, ret =strncpy_from_user (dst, src, maxlen);if (ret == maxlen) - dst[--ret] ='\0' ; + dst[ret -1 ] ='\0' ; +else if (ret >=0 ) +/* +* Include the terminating null byte. In this case it +* was copied by strncpy_from_user but not accounted +* for in ret. +*/ + ret++;if (ret <0 ) {/* Failed to fetch string */ ((u8 *)get_rloc_data (dest))[0 ] ='\0' ;Signature · 1 line
Masami Hiramatsu <[email protected]>
Signature · 1 line
Masami Hiramatsu <[email protected]>
Re: [PATCH stable-4.19 v2] tracing/uprobes: Fix output for multiple string arguments
In reply to Masami HiramatsuQuoted 65 lines · Masami Hiramatsu
From: Andreas Ziegler <[email protected]>
commit
0722069a5374b904ec1a67f91249f90e1cfae259upstream.When printing multiple uprobe arguments as strings the output for the earlier arguments would also include all later string arguments.
This is best explained in an example:
Consider adding a uprobe to a function receiving two strings as parameters which is at offset
0xa0in strlib.so and we want to print both parameters when the uprobe is hit (onx86_64):$ echo 'p:func /lib/strlib.so:0xa0 +0(%di):string +0(%si):string' > \ /sys/kernel/debug/tracing/uprobe_eventsWhen the function is called as
func("foo", "bar")and we hit the probe, the trace file shows a line like the following:[...] func: (
0x7f7e683706a0) arg1="foobar" arg2="bar"Note the extra "bar" printed as part of arg1. This behaviour stacks up for additional string arguments.
The strings are stored in a dynamically growing part of the uprobe buffer by
fetch_store_string()after copying them from userspace viastrncpy_from_user(). The return value ofstrncpy_from_user()is then directly used as the required size for the string. However, this does not take the terminating null byte into account as the documentation forstrncpy_from_user()cleary states that it "[...] returns the length of the string (not including the trailing NUL)" even though the null byte will be copied to the destination.Therefore, subsequent calls to
fetch_store_string()will overwrite the terminating null byte of the most recently fetched string with the first character of the current string, leading to the "accumulation" of strings in earlier arguments in the output.Fix this by incrementing the return value of
strncpy_from_user()by one if we did not hit the maximum buffer size.Cc Ingo Molnar <[email protected]>Fixes 5baaa59ef09e ("tracing/probes: Implement 'memory' fetch method for uprobes")Acked-by Masami Hiramatsu <[email protected]>Signed-off-by Andreas Ziegler <[email protected]>Signed-off-by Steven Rostedt (VMware) <[email protected]>Signed-off-by Masami Hiramatsu <[email protected]>--- kernel/trace/trace_uprobe.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-)kernel/trace/trace_uprobe.c +1
-1@@ -141,7 +141,14 @@ static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, ret =strncpy_from_user (dst, src, maxlen);if (ret == maxlen) - dst[--ret] ='\0' ; + dst[ret -1 ] ='\0' ;
kernel/trace/trace_uprobe.c static void FETCH_FUNC_NAME(memory, string)(struct pt_regs *regs, · line 154
Quoted 10 lines
kernel/trace/trace_uprobe.c +7
-0+else if (ret >=0 ) +/* +* Include the terminating null byte. In this case it +* was copied by strncpy_from_user but not accounted +* for in ret. +*/ + ret++;if (ret <0 ) {/* Failed to fetch string */ ((u8 *)get_rloc_data (dest))[0 ] ='\0' ;
All now queued up, thanks.
greg k-h