NFSD: Replace all non-returning strlcpy with strscpy
Subsystems (2)
PATCH 1/1 NFSD: Replace all non-returning strlcpy with strscpy
strlcpy() reads the entire source buffer first. This read may exceed the destination size limit. This is both inefficient and can lead to linear read overflows if a source string is not NUL-terminated [1]. In an effort to remove strlcpy() completely [2], replace strlcpy() here with strscpy().
No return values were used, so direct replacement is safe.
[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89
--- fs/nfsd/trace.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
fs/nfsd/trace.h +1 -1
@@ -1370,7 +1370,7 @@ TRACE_EVENT(nfsd_cb_setup,TP_fast_assign ( __entry->cl_boot = clp->cl_clientid.cl_boot; __entry->cl_id = clp->cl_clientid.cl_id; -strlcpy (__entry->netid, netid,sizeof (__entry->netid)); +strscpy (__entry->netid, netid,sizeof (__entry->netid)); __entry->authflavor = authflavor;__assign_sockaddr (addr, &clp->cl_cb_conn.cb_addr, clp->cl_cb_conn.cb_addrlen)
Hello Azeem -
Quoted 9 lines · Azeem Shaikh
strlcpy()reads the entire source buffer first. This read may exceed the destination size limit. This is both inefficient and can lead to linear read overflows if a source string is not NUL-terminated [1]. In an effort to removestrlcpy()completely [2], replacestrlcpy()here withstrscpy().
No return values were used, so direct replacement is safe.
Actually netid should use the __string() and __assign_str() macros rather than open-coding a string copy, I think.
Quoted 3 lines
[1] https://www.kernel.org/doc/html/latest/process/deprecated.html#strlcpy
[2] https://github.com/KSPP/linux/issues/89
I suggest adding:
Quoted 10 lines
Signed-off-by Azeem Shaikh <[email protected]>---
fs/nfsd/trace.h| 2 +-1 file changed, 1 insertion(+), 1 deletion(-)fs/nfsd/trace.h +0
-0@@ -1370,7 +1370,7 @@ TRACE_EVENT(nfsd_cb_setup,
Quoted 9 lines
TP_fast_assign(
__entry->cl_boot=clp->cl_clientid.cl_boot;__entry->cl_id=clp->cl_clientid.cl_id;
-strlcpy(__entry->netid, netid, sizeof(__entry->netid));
+strscpy(__entry->netid, netid, sizeof(__entry->netid));__entry->authflavor= authflavor;__assign_sockaddr(addr, &clp->cl_cb_conn.cb_addr, clp->cl_cb_conn.cb_addrlen)
--
Chuck Lever
Quoted 12 lines · Chuck Lever III
Hello Azeem -
Quoted 9 lines · Azeem Shaikh
strlcpy()reads the entire source buffer first. This read may exceed the destination size limit. This is both inefficient and can lead to linear read overflows if a source string is not NUL-terminated [1]. In an effort to removestrlcpy()completely [2], replacestrlcpy()here withstrscpy().
No return values were used, so direct replacement is safe.
Quoted 2 lines
Actually netid should use the
__string()and__assign_str()macros rather than open-coding a string copy, I think.
Ah, hm, yeah, this is tracing wrappers.
Steve, is there a reason __assign_str() is using "strcpy" and not strscpy()?
Quoted 1 line
Fixes 3c92fba557c6 ("NFSD: Enhance the nfsd_cb_setup tracepoint")
Yeah, that works. I was on the fence about adding Fixes for these kinds of refactoring. Like, it's not really _broken_; we're just trying to remove the API.
Quoted 5 lines
Quoted 5 lines
Signed-off-by Azeem Shaikh <[email protected]>---
fs/nfsd/trace.h| 2 +-1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 13 lines
Quoted 5 lines
fs/nfsd/trace.h +0
-0@@ -1370,7 +1370,7 @@ TRACE_EVENT(nfsd_cb_setup,Quoted 8 lines
TP_fast_assign(
__entry->cl_boot=clp->cl_clientid.cl_boot;__entry->cl_id=clp->cl_clientid.cl_id;
-strlcpy(__entry->netid, netid, sizeof(__entry->netid));
+strscpy(__entry->netid, netid, sizeof(__entry->netid));__entry->authflavor= authflavor;__assign_sockaddr(addr, &clp->cl_cb_conn.cb_addr, clp->cl_cb_conn.cb_addrlen)
Leaving code context for Steve to see...
Signature · 1 line
Kees Cook
Thanks Chuck and Kees for the review.
Quoted 2 lines
Quoted 2 lines
Actually netid should use the
__string()and__assign_str()macros rather than open-coding a string copy, I think.
Do you mean something like this?
fs/nfsd/trace.h +3 -3
@@ -1365,19 +1365,19 @@ TRACE_EVENT(nfsd_cb_setup, __field(u32, cl_id) __field(unsigned long, authflavor) __sockaddr(addr, clp->cl_cb_conn.cb_addrlen) - __array(unsigned char, netid, 8) + __string(netid, netid) ),TP_fast_assign ( __entry->cl_boot = clp->cl_clientid.cl_boot; __entry->cl_id = clp->cl_clientid.cl_id; - strlcpy(__entry->netid, netid, sizeof(__entry->netid)); +__assign_str (netid, netid); __entry->authflavor = authflavor;__assign_sockaddr (addr, &clp->cl_cb_conn.cb_addr, clp->cl_cb_conn.cb_addrlen) ),TP_printk ("addr=%pISpc client %08x:%08x proto=%s flavor=%s" ,__get_sockaddr (addr), __entry->cl_boot, __entry->cl_id, - __entry->netid, show_nfsd_authflavor(__entry->authflavor)) +__get_str (netid),show_nfsd_authflavor (__entry->authflavor)) );
Quoted 1 line
Quoted 1 line
Fixes 3c92fba557c6 ("NFSD: Enhance the nfsd_cb_setup tracepoint")
Ack.
Quoted 26 lines · Azeem Shaikh
Thanks Chuck and Kees for the review.
Quoted 2 lines
Quoted 2 lines
Actually netid should use the
__string()and__assign_str()macros rather than open-coding a string copy, I think.Do you mean something like this?
fs/nfsd/trace.h +1
-2@@ -1365,19 +1365,19 @@ TRACE_EVENT(nfsd_cb_setup, __field(u32, cl_id) __field(unsigned long, authflavor) __sockaddr(addr, clp->cl_cb_conn.cb_addrlen) - __array(unsigned char, netid, 8) + __string(netid, netid) ),TP_fast_assign ( __entry->cl_boot = clp->cl_clientid.cl_boot; __entry->cl_id = clp->cl_clientid.cl_id; - strlcpy(__entry->netid, netid, sizeof(__entry->netid));
fs/nfsd/trace.h TRACE_EVENT(nfsd_cb_setup, · line 1381
Quoted 10 lines
fs/nfsd/trace.h +2
-1+__assign_str (netid, netid); __entry->authflavor = authflavor;__assign_sockaddr (addr, &clp->cl_cb_conn.cb_addr, clp->cl_cb_conn.cb_addrlen) ),TP_printk ("addr=%pISpc client %08x:%08x proto=%s flavor=%s" ,__get_sockaddr (addr), __entry->cl_boot, __entry->cl_id, - __entry->netid,show_nfsd_authflavor (__entry->authflavor)) +__get_str (netid),show_nfsd_authflavor (__entry->authflavor)) );
Yes. I don't remember why I chose to open-code the string handling in here.
I agree with Kees, also: __assign_str() needs to handle the memory accesses properly, and I had assumed it did that already. My bad. That IMO should be handled as a separate patch.
Quoted 3 lines
fs/nfsd/trace.h +0
-0>> Fixes: 3c92fba557c6 ("NFSD: Enhance the nfsd_cb_setup tracepoint")Ack.
I agree that it's not a high priority backport. However, I expect those who want to use trace points on KASAN- enabled stable kernels might prefer to have the __assign_str fix in place, along with this one.
So, I'm good with including Fixes: or leaving it off. Up to you!
--
Chuck Lever
Quoted 41 lines · Chuck Lever III
Quoted 27 lines · Azeem Shaikh
Thanks Chuck and Kees for the review.
Quoted 2 lines
Quoted 2 lines
Actually netid should use the
__string()and__assign_str()macros rather than open-coding a string copy, I think.Do you mean something like this?
fs/nfsd/trace.h +1
-2@@ -1365,19 +1365,19 @@ TRACE_EVENT(nfsd_cb_setup, __field(u32, cl_id) __field(unsigned long, authflavor) __sockaddr(addr, clp->cl_cb_conn.cb_addrlen) - __array(unsigned char, netid, 8) + __string(netid, netid) ),TP_fast_assign ( __entry->cl_boot = clp->cl_clientid.cl_boot; __entry->cl_id = clp->cl_clientid.cl_id; - strlcpy(__entry->netid, netid, sizeof(__entry->netid));fs/nfsd/trace.h TRACE_EVENT(nfsd_cb_setup, · line 1381
Quoted 10 lines
fs/nfsd/trace.h +2
-1+__assign_str (netid, netid); __entry->authflavor = authflavor;__assign_sockaddr (addr, &clp->cl_cb_conn.cb_addr, clp->cl_cb_conn.cb_addrlen) ),TP_printk ("addr=%pISpc client %08x:%08x proto=%s flavor=%s" ,__get_sockaddr (addr), __entry->cl_boot, __entry->cl_id, - __entry->netid,show_nfsd_authflavor (__entry->authflavor)) +__get_str (netid),show_nfsd_authflavor (__entry->authflavor)) );Yes. I don't remember why I chose to open-code the string handling in here.
Quoted 4 lines
I agree with Kees, also:
__assign_str()needs to handle the memory accesses properly, and I had assumed it did that already. My bad. That IMO should be handled as a separate patch.
Thanks I'll send out a patch with the __assign_str replacement for now. I'll wait for Kees and Steve to respond on how to fix the memory access in __assign_str.
Quoted 4 lines
Quoted 3 lines
fs/nfsd/trace.h +0
-0>> Fixes: 3c92fba557c6 ("NFSD: Enhance the nfsd_cb_setup tracepoint")Ack.
Quoted 8 lines
I agree that it's not a high priority backport. However, I expect those who want to use trace points on KASAN- enabled stable kernels might prefer to have the
__assign_strfix in place, along with this one.So, I'm good with including Fixes: or leaving it off. Up to you!
Will include it :)
Quoted 13 lines · On Thu, 11 May 2023 09:32:31 -0700 Kees Cook <[email protected]> wrote:
Quoted 12 lines · Chuck Lever III
Hello Azeem -
Quoted 9 lines · Azeem Shaikh
strlcpy()reads the entire source buffer first. This read may exceed the destination size limit. This is both inefficient and can lead to linear read overflows if a source string is not NUL-terminated [1]. In an effort to removestrlcpy()completely [2], replacestrlcpy()here withstrscpy().
No return values were used, so direct replacement is safe.
Quoted 7 lines
Quoted 2 lines
Actually netid should use the
__string()and__assign_str()macros rather than open-coding a string copy, I think.Ah, hm, yeah, this is tracing wrappers.
Steve, is there a reason
__assign_str()is using "strcpy" and notstrscpy()?
Yes. Because __assign_str() predates strscpy() ;-)
Note, to use __assign_str(), you first need to do __string(), which will do (in all that TRACE_EVENT() macro magic):
#undef __dynamic_array
#define __dynamic_array(type, item, len) \
__item_length = (len) * sizeof(type); \
__data_offsets->item = __data_size + \
offsetof(typeof(*entry), __data); \
__data_offsets->item |= __item_length << 16; \
__data_size += __item_length;#undef __string
#define __string(item, src) __dynamic_array(char, item, \
strlen((src) ? (const char *)(src) : "(null)") + 1)In order to save a dynamic size string (or array) it must first calculate that size with a strlen(). If the source is not terminated, then this will crash there regardless.
The strlen() is to know how much to allocate on the ring buffer, then a copy is done to copy it. I guess you could be worried about the size "changing" between the time it is allocated and copied. If that's the concern then we could do a length copy.
Quoted 12 lines
Quoted 2 lines
Fixes 3c92fba557c6 ("NFSD: Enhance the nfsd_cb_setup tracepoint")Yeah, that works. I was on the fence about adding Fixes for these kinds of refactoring. Like, it's not really _broken_; we're just trying to remove the API.
Quoted 5 lines
Quoted 5 lines
Signed-off-by Azeem Shaikh <[email protected]>---
fs/nfsd/trace.h| 2 +-1 file changed, 1 insertion(+), 1 deletion(-)
Quoted 16 lines
Quoted 5 lines
Quoted 5 lines
fs/nfsd/trace.h +0
-0@@ -1370,7 +1370,7 @@ TRACE_EVENT(nfsd_cb_setup,Quoted 8 lines
Quoted 8 lines
TP_fast_assign(
__entry->cl_boot=clp->cl_clientid.cl_boot;__entry->cl_id=clp->cl_clientid.cl_id;
-strlcpy(__entry->netid, netid, sizeof(__entry->netid));
+strscpy(__entry->netid, netid, sizeof(__entry->netid));__entry->authflavor= authflavor;__assign_sockaddr(addr, &clp->cl_cb_conn.cb_addr, clp->cl_cb_conn.cb_addrlen)Leaving code context for Steve to see...
The above isn't __assign_str(). Not exactly sure what you want me to see here.
-- Steve