sunrpc: on successful gss error pipe write, don't return error (try #2)
Subsystems (3)
PATCH 1/1 sunrpc: on successful gss error pipe write, don't return error (try #2)
When handling the gssd downcall, the kernel should distinguish between a successful downcall that contains an error code and a failed downcall (i.e. where the parsing failed or some other sort of problem occurred).
In the former case, gss_pipe_downcall should be returning the number of bytes written to the pipe instead of an error.
--- net/sunrpc/auth_gss/auth_gss.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)
net/sunrpc/auth_gss/auth_gss.c +3 -0
@@ -645,6 +645,9 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)if (IS_ERR (p)) { err =PTR_ERR (p); gss_msg->msg.errno = (err == -EAGAIN ) ? -EAGAIN : -EACCES ; +/* special case: downcall was successful, but held an error */ +if (err == -EACCES ) + err = mlen; goto err_release_msg; } gss_msg->ctx =gss_get_ctx (ctx);
Signature · 6 lines
1.6.5.2 _______________________________________________ NFSv4 mailing list [email protected] http://linux-nfs.org/cgi-bin/mailman/listinfo/nfsv4
Quoted 17 lines · Jeff Layton
When handling the gssd downcall, the kernel should distinguish between a successful downcall that contains an error code and a failed downcall (i.e. where the parsing failed or some other sort of problem occurred).
In the former case,
gss_pipe_downcallshould be returning the number of bytes written to the pipe instead of an error.Signed-off-by Jeff Layton <[email protected]>--- net/sunrpc/auth_gss/auth_gss.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)net/sunrpc/auth_gss/auth_gss.c +0
-0
net/sunrpc/auth_gss/auth_gss.c gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen) · line 650
Quoted 6 lines
net/sunrpc/auth_gss/auth_gss.c +3
-0@@ -645,6 +645,9 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)if (IS_ERR (p)) { err =PTR_ERR (p); gss_msg->msg.errno = (err == -EAGAIN ) ? -EAGAIN : -EACCES ; +/* special case: downcall was successful, but held an error */ +if (err == -EACCES ) + err = mlen;
That line immediately above your fix still looks wrong. The point is that AFAICS, err is never going to be set to EAGAIN. It can be EFAULT, ENOSYS, or ENOMEM, but it will never be EAGAIN...
I think we should rather reverse that test. Really, what we want to do, is to set msg.errno to -EAGAIN for -EFAULT and -ENOMEM (and probably for ENOSYS too), and then set it to -EACCES _only_ in the case where the user was not authorised.
Trond
Quoted 29 lines · On Fri, 18 Dec 2009 14:05:39 -0500 Trond Myklebust…
Quoted 17 lines · Jeff Layton
When handling the gssd downcall, the kernel should distinguish between a successful downcall that contains an error code and a failed downcall (i.e. where the parsing failed or some other sort of problem occurred).
In the former case,
gss_pipe_downcallshould be returning the number of bytes written to the pipe instead of an error.Signed-off-by Jeff Layton <[email protected]>--- net/sunrpc/auth_gss/auth_gss.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)net/sunrpc/auth_gss/auth_gss.c +0
-0net/sunrpc/auth_gss/auth_gss.c gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen) · line 650
Quoted 6 lines
net/sunrpc/auth_gss/auth_gss.c +3
-0@@ -645,6 +645,9 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)if (IS_ERR (p)) { err =PTR_ERR (p); gss_msg->msg.errno = (err == -EAGAIN ) ? -EAGAIN : -EACCES ; +/* special case: downcall was successful, but held an error */ +if (err == -EACCES ) + err = mlen;That line immediately above your fix still looks wrong. The point is that AFAICS, err is never going to be set to EAGAIN. It can be EFAULT, ENOSYS, or ENOMEM, but it will never be EAGAIN...
Quoted 4 lines
I think we should rather reverse that test. Really, what we want to do, is to set msg.errno to
-EAGAINfor-EFAULTand-ENOMEM(and probably for ENOSYS too), and then set it to-EACCES_only_ in the case where the user was not authorised.
What should we do if err is "none of the above"? Set msg.errno to -EACCES and return the error to the pipe writer?
Signature · 1 line
Jeff Layton <[email protected]>
Quoted 32 lines · Jeff Layton
Quoted 29 lines · On Fri, 18 Dec 2009 14:05:39 -0500 Trond Myklebust…
Quoted 17 lines · Jeff Layton
When handling the gssd downcall, the kernel should distinguish between a successful downcall that contains an error code and a failed downcall (i.e. where the parsing failed or some other sort of problem occurred).
In the former case,
gss_pipe_downcallshould be returning the number of bytes written to the pipe instead of an error.Signed-off-by Jeff Layton <[email protected]>--- net/sunrpc/auth_gss/auth_gss.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)net/sunrpc/auth_gss/auth_gss.c +0
-0net/sunrpc/auth_gss/auth_gss.c gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen) · line 650
Quoted 6 lines
net/sunrpc/auth_gss/auth_gss.c +3
-0@@ -645,6 +645,9 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)if (IS_ERR (p)) { err =PTR_ERR (p); gss_msg->msg.errno = (err == -EAGAIN ) ? -EAGAIN : -EACCES ; +/* special case: downcall was successful, but held an error */ +if (err == -EACCES ) + err = mlen;That line immediately above your fix still looks wrong. The point is that AFAICS, err is never going to be set to EAGAIN. It can be EFAULT, ENOSYS, or ENOMEM, but it will never be EAGAIN...
Quoted 8 lines
Quoted 4 lines
I think we should rather reverse that test. Really, what we want to do, is to set msg.errno to
-EAGAINfor-EFAULTand-ENOMEM(and probably for ENOSYS too), and then set it to-EACCES_only_ in the case where the user was not authorised.What should we do if err is "none of the above"? Set msg.errno to
-EACCESand return the error to the pipe writer?
The question is will it ever be 'none of the above'? We clearly cannot be returning arbitrary errors to gssd, so we need to define a set that makes sense. The only other error I can see that we might to add to the above list, would be EINVAL (to mean 'you just sent me some garbage argument that I cannot decode').
Retrying the upcall would seem to be the correct thing to do in case of EINVAL too...
Trond
Quoted 42 lines · On Fri, 18 Dec 2009 15:24:26 -0500 Trond Myklebust…
Quoted 32 lines · Jeff Layton
Quoted 29 lines · On Fri, 18 Dec 2009 14:05:39 -0500 Trond Myklebust…
Quoted 17 lines · Jeff Layton
When handling the gssd downcall, the kernel should distinguish between a successful downcall that contains an error code and a failed downcall (i.e. where the parsing failed or some other sort of problem occurred).
In the former case,
gss_pipe_downcallshould be returning the number of bytes written to the pipe instead of an error.Signed-off-by Jeff Layton <[email protected]>--- net/sunrpc/auth_gss/auth_gss.c | 3 +++ 1 files changed, 3 insertions(+), 0 deletions(-)net/sunrpc/auth_gss/auth_gss.c +0
-0net/sunrpc/auth_gss/auth_gss.c gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen) · line 650
Quoted 6 lines
net/sunrpc/auth_gss/auth_gss.c +3
-0@@ -645,6 +645,9 @@ gss_pipe_downcall(struct file *filp, const char __user *src, size_t mlen)if (IS_ERR (p)) { err =PTR_ERR (p); gss_msg->msg.errno = (err == -EAGAIN ) ? -EAGAIN : -EACCES ; +/* special case: downcall was successful, but held an error */ +if (err == -EACCES ) + err = mlen;That line immediately above your fix still looks wrong. The point is that AFAICS, err is never going to be set to EAGAIN. It can be EFAULT, ENOSYS, or ENOMEM, but it will never be EAGAIN...
Quoted 8 lines
Quoted 4 lines
I think we should rather reverse that test. Really, what we want to do, is to set msg.errno to
-EAGAINfor-EFAULTand-ENOMEM(and probably for ENOSYS too), and then set it to-EACCES_only_ in the case where the user was not authorised.What should we do if err is "none of the above"? Set msg.errno to
-EACCESand return the error to the pipe writer?
Quoted 9 lines
The question is will it ever be 'none of the above'? We clearly cannot be returning arbitrary errors to gssd, so we need to define a set that makes sense. The only other error I can see that we might to add to the above list, would be EINVAL (to mean 'you just sent me some garbage argument that I cannot decode').
Retrying the upcall would seem to be the correct thing to do in case of EINVAL too...
Well, I'd like to think that we'd always get the lower layers right, but I know better :). We probably ought to code defensively here and account for the possibility that this function could return a different error due to an inadvertent change in a lower function.
If the "bad" error is transient then -EAGAIN would probably just paper over it. If it's not transient then the mount would hang. -EACCES is a nice hard error, but could be confused with a "real" -EACCES. We could BUG() in that case if we're sure it'd be a real bug...
Signature · 1 line
Jeff Layton <[email protected]>
Quoted 9 lines · Jeff Layton
Well, I'd like to think that we'd always get the lower layers right, but I know better :). We probably ought to code defensively here and account for the possibility that this function could return a different error due to an inadvertent change in a lower function.
If the "bad" error is transient then
-EAGAINwould probably just paper over it. If it's not transient then the mount would hang.-EACCESis a nice hard error, but could be confused with a "real"-EACCES. We couldBUG()in that case if we're sure it'd be a real bug...
switch (err) {
case -EACCES:gss_msg->msg.errno = -EACCES;
break;
case -EFAULT:
case -ENOMEM:
case -EINVAL:
case -ENOSYS:gss_msg->msg.errno = -EAGAIN;
break;
default:BUG();
}
Yup. That looks good to me... ..and with the fixes to gss_import_sec_context_kerberos and gss_import_sec_context, it will actually be correct too.
BTW: get_key() returns EINVAL on error, so that does indeed need to be supported in the above list.
Trond
Quoted 14 lines · Trond Myklebust
switch (err) {
case-EACCES:gss_msg->msg.errno=-EACCES;
break;
case-EFAULT:
case-ENOMEM:
case-EINVAL:
case-ENOSYS:gss_msg->msg.errno=-EAGAIN;
break;
default:BUG();
}
Quoted 1 line
Yup. That looks good to me...
Well... Minus the helpful reformatting that evolution did for me after I pressed the 'send' button.
Trond