On 14/10/2010 19:10, Roland McGrath wrote: >> Shouldn't we change io_write() too? > Yes, they are exact parallels. Here is a patch that should make io_read() and io_write() return ESPIPE when called with offset != -1 for non-seekable objects. I've tested all the components affected by this patch, and AFAICS everything is working fine (and now pread() behaves as expected).
I've included a change for the documentation as well. As I'm not a
native speaker, my wording might not be elegant: someone might
want to proofread it.
diff --git a/boot/boot.c b/boot/boot.c
index 1a4e1b1..cc62b60 100644
--- a/boot/boot.c
+++ b/boot/boot.c
@@ -1421,6 +1421,8 @@ S_io_write (mach_port_t object,
{
if (object != pseudo_console)
return EOPNOTSUPP;
+ else if (offset != (off_t) -1)
+ return ESPIPE;
#if 0
if (console_send_rights)
@@ -1448,6 +1450,8 @@ S_io_read (mach_port_t object,
if (object != pseudo_console)
return EOPNOTSUPP;
+ if (offset != (off_t) -1)
+ return ESPIPE;
#if 0
if (console_send_rights)
diff --git a/console-client/kbd-repeat.c b/console-client/kbd-repeat.c
index 4bf9268..75b773f 100644
--- a/console-client/kbd-repeat.c
+++ b/console-client/kbd-repeat.c
@@ -131,6 +131,8 @@ repeater_read (struct protid *cred, char **data,
return EOPNOTSUPP;
else if (! (cred->po->openstat & O_READ))
return EBADF;
+ else if (offset != (off_t) -1)
+ return ESPIPE;
mutex_lock (&global_lock);
while (amount > kbdbuf.size)
diff --git a/console-client/pc-mouse.c b/console-client/pc-mouse.c
index cf8987b..426817d 100644
--- a/console-client/pc-mouse.c
+++ b/console-client/pc-mouse.c
@@ -174,7 +174,9 @@ repeater_read (struct protid *cred, char **data,
return EOPNOTSUPP;
else if (! (cred->po->openstat & O_READ))
return EBADF;
-
+ else if (offset != (off_t) -1)
+ return ESPIPE;
+
mutex_lock (&global_lock);
while (!mousebuf.size)
{
diff --git a/console/console.c b/console/console.c
index dad7516..606f646 100644
--- a/console/console.c
+++ b/console/console.c
@@ -1114,13 +1114,18 @@ netfs_attempt_read (struct iouser *cred, struct node
*np,
mutex_unlock (&np->lock);
if (np == vcons->cons_node)
{
- ssize_t amt = input_dequeue (vcons->input,
- /* cred->po->openstat & O_NONBLOCK */ 0,
- data, *len);
- if (amt == -1)
- err = errno;
- else
- *len = amt;
+ if (offset != (off_t) -1)
+ err = ESPIPE;
+ else
+ {
+ ssize_t amt = input_dequeue (vcons->input,
+ /* cred->po->openstat & O_NONBLOCK */ 0,
+ data, *len);
+ if (amt == -1)
+ err = errno;
+ else
+ *len = amt;
+ }
}
else
{
diff --git a/doc/hurd.texi b/doc/hurd.texi
index 098bebf..5a60d78 100644
--- a/doc/hurd.texi
+++ b/doc/hurd.texi
@@ -1649,17 +1649,18 @@ ID's.
@subsection Simple Operations
@findex io_write
-Users write to I/O ports by calling the @code{io_write} RPC. They
-specify an @var{offset} parameter; if the object supports writing at
-arbitrary offsets, the server should honour this parameter. If @math{-1}
-is passed as the offset, then the server should use the default file
-pointer. The server should return the amount of data which was
-successfully written. If the operation was interrupted after some but
-not all of the data was written, then it is considered to have succeeded
-and the server should return the amount written. If the port is not an
-I/O port at all, the server should reply with the error
-...@code{eopnotsupp}. If the port is an I/O port, but does not happen to
-support writing, then the correct error is @code{EBADF}.
+Users write to I/O ports by calling the @code{io_write} RPC. They specify an
+...@var{offset} parameter; if the object supports writing at arbitrary
offsets, the
+server should honour this parameter. If @math{-1} is passed as the offset, then
+the server should use the default file pointer. For objects that don't support
+writing at arbitrary offsets, the offset should always be @math{-1}, or the
+server should reply with the error @code{ESPIPE}. The server should return the
+amount of data which was successfully written. If the operation was interrupted
+after some but not all of the data was written, then it is considered to have
+succeeded and the server should return the amount written. If the port is not
an
+I/O port at all, the server should reply with the error @code{EOPNOTSUPP}. If
+the port is an I/O port, but does not happen to support writing, then the
+correct error is @code{EBADF}.
@findex io_read
Users read from I/O ports by calling the @code{io_read} RPC. They
@@ -1677,12 +1678,12 @@ then the server may return @code{EINTR} rather than
actually filling the
buffer (taking care that any modifications of the default file pointer
have been reversed). Preferably, however, servers should return data.
-There are two categories of objects: seekable and non-seekable.
-Seekable objects must accept arbitrary offset parameters in the
-...@code{io_read} and @code{io_write} calls, and must implement the
-...@code{io_seek} call. Non-seekable objects must ignore the offset
-parameters to @code{io_read} and @code{io_write}, and should return
-...@code{espipe} to the @code{io_seek} call.
+There are two categories of objects: seekable and non-seekable. Seekable
objects
+must accept arbitrary offset parameters in the @code{io_read} and
+...@code{io_write} calls, and must implement the @code{io_seek} call. For
+non-seekable objects, the offset parameter to @code{io_read} and
@code{io_write}
+should always be @math{-1}, or @code{ESPIPE} should be returned; they should
+also return @code{ESPIPE} to the @code{io_seek} call.
@c FIXME: should that last `should' be replaced with `must'?
@c tb: maybe, but perhaps not. There might be a reason to implement a
diff --git a/hurd/io.defs b/hurd/io.defs
index d30233e..06855ce 100644
--- a/hurd/io.defs
+++ b/hurd/io.defs
@@ -32,13 +32,13 @@ IO_IMPORTS
INTR_INTERFACE
-/* Write data to an IO object. If offset is -1, write at the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount successfully written is returned in AMOUNT. A
- given user should not have more than one outstanding io_write on an
- object at a time; servers implement congestion control by delaying
- responses to io_write. Servers may drop data (returning ENOBUFS)
- if they recevie more than one write when not prepared for it. */
+/* Write data to an IO object. If offset is -1, write at the object maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount successfully written is returned in AMOUNT. A given
user
+ should not have more than one outstanding io_write on an object at a time;
+ servers implement congestion control by delaying responses to io_write.
+ Servers may drop data (returning ENOBUFS) if they recevie more than one
write
+ when not prepared for it. */
routine io_write (
io_object: io_t;
RPT
@@ -46,9 +46,9 @@ routine io_write (
offset: loff_t;
out amount: vm_size_t);
-/* Read data from an IO object. If offset if -1, read from the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount desired to be read is in AMOUNT. */
+/* Read data from an IO object. If offset if -1, read from the object
maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount desired to be read is in AMOUNT. */
routine io_read (
io_object: io_t;
RPT
diff --git a/pfinet/io-ops.c b/pfinet/io-ops.c
index ef8d851..554025b 100644
--- a/pfinet/io-ops.c
+++ b/pfinet/io-ops.c
@@ -47,6 +47,8 @@ S_io_write (struct sock_user *user,
if (!user)
return EOPNOTSUPP;
+ else if (offset != (off_t) -1)
+ return ESPIPE;
__mutex_lock (&global_lock);
become_task (user);
@@ -81,6 +83,8 @@ S_io_read (struct sock_user *user,
if (!user)
return EOPNOTSUPP;
+ if (offset != (off_t) -1)
+ return ESPIPE;
/* Instead of this, we should peek and the socket and only
allocate as much as necessary. */
diff --git a/pfinet/tunnel.c b/pfinet/tunnel.c
index c4f9580..a4a69f8 100644
--- a/pfinet/tunnel.c
+++ b/pfinet/tunnel.c
@@ -271,9 +271,9 @@ error_t (*trivfs_check_open_hook)(struct trivfs_control *,
is about to be destroyed. */
void (*trivfs_protid_destroy_hook) (struct trivfs_protid *) = pi_destroy_hook;
-/* Read data from an IO object. If offset is -1, read from the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount desired to be read is in AMOUNT. */
+/* Read data from an IO object. If offset is -1, read from the object
maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount desired to be read is in AMOUNT. */
error_t
trivfs_S_io_read (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t reply_type,
@@ -288,6 +288,8 @@ trivfs_S_io_read (struct trivfs_protid *cred,
return EOPNOTSUPP;
else if (! (cred->po->openmodes & O_READ))
return EBADF;
+ else if (offs != (off_t) -1)
+ return ESPIPE;
if (cred->pi.class != tunnel_class)
return EOPNOTSUPP;
@@ -345,13 +347,13 @@ trivfs_S_io_read (struct trivfs_protid *cred,
return 0;
}
-/* Write data to an IO object. If offset is -1, write at the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount successfully written is returned in amount. A
- given user should not have more than one outstanding io_write on an
- object at a time; servers implement congestion control by delaying
- responses to io_write. Servers may drop data (returning ENOBUFS)
- if they receive more than one write when not prepared for it. */
+/* Write data to an IO object. If offset is -1, write at the object maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount successfully written is returned in amount. A given
user
+ should not have more than one outstanding io_write on an object at a time;
+ servers implement congestion control by delaying responses to io_write.
+ Servers may drop data (returning ENOBUFS) if they receive more than one
write
+ when not prepared for it. */
error_t
trivfs_S_io_write (struct trivfs_protid *cred,
mach_port_t reply,
@@ -369,6 +371,8 @@ trivfs_S_io_write (struct trivfs_protid *cred,
return EOPNOTSUPP;
else if (! (cred->po->openmodes & O_WRITE))
return EBADF;
+ else if (offset != (off_t) -1)
+ return ESPIPE;
if (cred->pi.class != tunnel_class)
return EOPNOTSUPP;
diff --git a/pflocal/io.c b/pflocal/io.c
index 74c2f97..d02697d 100644
--- a/pflocal/io.c
+++ b/pflocal/io.c
@@ -39,9 +39,9 @@
#include "io_S.h"
#include "interrupt_S.h"
-/* Read data from an IO object. If offset if -1, read from the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount desired to be read is in amount. */
+/* Read data from an IO object. If offset if -1, read from the object
maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount desired to be read is in amount. */
error_t
S_io_read (struct sock_user *user,
char **data, mach_msg_type_number_t *data_len,
@@ -52,6 +52,8 @@ S_io_read (struct sock_user *user,
if (!user)
return EOPNOTSUPP;
+ if (offset != (off_t) -1)
+ return ESPIPE;
err = sock_acquire_read_pipe (user->sock, &pipe);
if (err == EPIPE)
@@ -71,13 +73,13 @@ S_io_read (struct sock_user *user,
return err;
}
-/* Write data to an IO object. If offset is -1, write at the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount successfully written is returned in amount. A
- given user should not have more than one outstanding io_write on an
- object at a time; servers implement congestion control by delaying
- responses to io_write. Servers may drop data (returning ENOBUFS)
- if they recevie more than one write when not prepared for it. */
+/* Write data to an IO object. If offset is -1, write at the object maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount successfully written is returned in amount. A given
user
+ should not have more than one outstanding io_write on an object at a time;
+ servers implement congestion control by delaying responses to io_write.
+ Servers may drop data (returning ENOBUFS) if they recevie more than one
write
+ when not prepared for it. */
error_t
S_io_write (struct sock_user *user,
char *data, mach_msg_type_number_t data_len,
@@ -88,6 +90,8 @@ S_io_write (struct sock_user *user,
if (!user)
return EOPNOTSUPP;
+ else if (offset != (off_t) -1)
+ return ESPIPE;
err = sock_acquire_write_pipe (user->sock, &pipe);
if (!err)
diff --git a/storeio/io.c b/storeio/io.c
index 508df77..89626b2 100644
--- a/storeio/io.c
+++ b/storeio/io.c
@@ -75,9 +75,9 @@ trivfs_S_io_map (struct trivfs_protid *cred,
}
}
-/* Read data from an IO object. If offset if -1, read from the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount desired to be read is in AMOUNT. */
+/* Read data from an IO object. If offset if -1, read from the object
maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount desired to be read is in AMOUNT. */
error_t
trivfs_S_io_read (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t reply_type,
@@ -113,13 +113,13 @@ trivfs_S_io_readable (struct trivfs_protid *cred,
}
}
-/* Write data to an IO object. If offset is -1, write at the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount successfully written is returned in amount. A
- given user should not have more than one outstanding io_write on an
- object at a time; servers implement congestion control by delaying
- responses to io_write. Servers may drop data (returning ENOBUFS)
- if they recevie more than one write when not prepared for it. */
+/* Write data to an IO object. If offset is -1, write at the object maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount successfully written is returned in amount. A given
user
+ should not have more than one outstanding io_write on an object at a time;
+ servers implement congestion control by delaying responses to io_write.
+ Servers may drop data (returning ENOBUFS) if they recevie more than one
write
+ when not prepared for it. */
error_t
trivfs_S_io_write (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t reply_type,
diff --git a/term/users.c b/term/users.c
index a4e110b..67c2446 100644
--- a/term/users.c
+++ b/term/users.c
@@ -580,6 +580,8 @@ trivfs_S_io_write (struct trivfs_protid *cred,
if (!cred)
return EOPNOTSUPP;
+ else if (offset != (off_t) -1)
+ return ESPIPE;
if (cred->pi.class == pty_class)
return pty_io_write (cred, data, datalen, amt);
@@ -658,6 +660,8 @@ trivfs_S_io_read (struct trivfs_protid *cred,
if (!cred)
return EOPNOTSUPP;
+ if (offset != (off_t) -1)
+ return ESPIPE;
if (cred->pi.class == pty_class)
return pty_io_read (cred, data, datalen, amount);
diff --git a/trans/fifo.c b/trans/fifo.c
index 39043ac..e49e57b 100644
--- a/trans/fifo.c
+++ b/trans/fifo.c
@@ -329,9 +329,9 @@ trivfs_S_io_map (struct trivfs_protid *cred,
/* ---------------------------------------------------------------- */
-/* Read data from an IO object. If offset if -1, read from the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount desired to be read is in AMT. */
+/* Read data from an IO object. If offset if -1, read from the object
maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount desired to be read is in AMT. */
error_t
trivfs_S_io_read (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t reply_type,
@@ -344,6 +344,8 @@ trivfs_S_io_read (struct trivfs_protid *cred,
err = EOPNOTSUPP;
else if (!(cred->po->openmodes & O_READ))
err = EBADF;
+ else if (offs != (off_t) -1)
+ err = ESPIPE;
else
{
struct pipe *pipe = cred->po->hook;
@@ -458,13 +460,13 @@ trivfs_S_io_select (struct trivfs_protid *cred,
/* ---------------------------------------------------------------- */
-/* Write data to an IO object. If offset is -1, write at the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount successfully written is returned in amount. A
- given user should not have more than one outstanding io_write on an
- object at a time; servers implement congestion control by delaying
- responses to io_write. Servers may drop data (returning ENOBUFS)
- if they recevie more than one write when not prepared for it. */
+/* Write data to an IO object. If offset is -1, write at the object maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount successfully written is returned in amount. A given
user
+ should not have more than one outstanding io_write on an object at a time;
+ servers implement congestion control by delaying responses to io_write.
+ Servers may drop data (returning ENOBUFS) if they recevie more than one
write
+ when not prepared for it. */
error_t
trivfs_S_io_write (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t reply_type,
@@ -475,6 +477,8 @@ trivfs_S_io_write (struct trivfs_protid *cred,
if (!cred)
err = EOPNOTSUPP;
+ else if (offs != (off_t) -1)
+ err = ESPIPE;
else
{
int flags = cred->po->openmodes;
diff --git a/trans/firmlink.c b/trans/firmlink.c
index 087e19d..551caba 100644
--- a/trans/firmlink.c
+++ b/trans/firmlink.c
@@ -197,9 +197,9 @@ trivfs_goaway (struct trivfs_control *cntl, int flags)
/* We store the file offset in po->hook (ick!) */
-/* Read data from an IO object. If offset if -1, read from the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount desired to be read is in AMT. */
+/* Read data from an IO object. If offset if -1, read from the object
maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount desired to be read is in AMT. */
error_t
trivfs_S_io_read (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t reply_type,
diff --git a/trans/hello-mt.c b/trans/hello-mt.c
index b933cfd..e80272c 100644
--- a/trans/hello-mt.c
+++ b/trans/hello-mt.c
@@ -117,9 +117,9 @@ close_hook (struct trivfs_peropen *peropen)
}
-/* Read data from an IO object. If offset is -1, read from the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount desired to be read is in AMOUNT. */
+/* Read data from an IO object. If offset is -1, read from the object
maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount desired to be read is in AMOUNT. */
error_t
trivfs_S_io_read (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t reply_type,
diff --git a/trans/hello.c b/trans/hello.c
index c49feeb..b6a49dd 100644
--- a/trans/hello.c
+++ b/trans/hello.c
@@ -107,9 +107,9 @@ close_hook (struct trivfs_peropen *peropen)
}
-/* Read data from an IO object. If offset is -1, read from the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount desired to be read is in AMOUNT. */
+/* Read data from an IO object. If offset is -1, read from the object
maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount desired to be read is in AMOUNT. */
error_t
trivfs_S_io_read (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t reply_type,
diff --git a/trans/new-fifo.c b/trans/new-fifo.c
index 5306dee..7ab9e6a 100644
--- a/trans/new-fifo.c
+++ b/trans/new-fifo.c
@@ -514,9 +514,9 @@ trivfs_S_io_map (struct trivfs_protid *cred,
/* ---------------------------------------------------------------- */
-/* Read data from an IO object. If offset if -1, read from the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount desired to be read is in AMT. */
+/* Read data from an IO object. If offset if -1, read from the object
maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount desired to be read is in AMT. */
error_t
trivfs_S_io_read (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t reply_type,
@@ -529,6 +529,8 @@ trivfs_S_io_read (struct trivfs_protid *cred,
err = EOPNOTSUPP;
else if (!(cred->po->openmodes & O_READ))
err = EBADF;
+ else if (offs != (off_t) -1)
+ err = ESPIPE;
else
{
struct pipe *pipe = cred->po->hook;
@@ -645,13 +647,13 @@ trivfs_S_io_select (struct trivfs_protid *cred,
/* ---------------------------------------------------------------- */
-/* Write data to an IO object. If offset is -1, write at the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount successfully written is returned in amount. A
- given user should not have more than one outstanding io_write on an
- object at a time; servers implement congestion control by delaying
- responses to io_write. Servers may drop data (returning ENOBUFS)
- if they recevie more than one write when not prepared for it. */
+/* Write data to an IO object. If offset is -1, write at the object maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount successfully written is returned in amount. A given
user
+ should not have more than one outstanding io_write on an object at a time;
+ servers implement congestion control by delaying responses to io_write.
+ Servers may drop data (returning ENOBUFS) if they recevie more than one
write
+ when not prepared for it. */
error_t
trivfs_S_io_write (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t reply_type,
@@ -664,6 +666,8 @@ trivfs_S_io_write (struct trivfs_protid *cred,
err = EOPNOTSUPP;
else if (!(cred->po->openmodes & O_WRITE))
err = EBADF;
+ else if (offs != (off_t) -1)
+ err = ESPIPE;
else
{
struct pipe *pipe = cred->po->hook;
diff --git a/trans/null.c b/trans/null.c
index 9673a75..440c196 100644
--- a/trans/null.c
+++ b/trans/null.c
@@ -133,9 +133,9 @@ trivfs_S_io_map (struct trivfs_protid *cred,
return EOPNOTSUPP; /* XXX should work! */
}
-/* Read data from an IO object. If offset if -1, read from the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount desired to be read is in AMT. */
+/* Read data from an IO object. If offset if -1, read from the object
maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
is
+ returned. The amount desired to be read is in AMT. */
kern_return_t
trivfs_S_io_read(struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t replytype,
@@ -204,13 +204,13 @@ trivfs_S_io_select (struct trivfs_protid *cred,
return 0;
}
-/* Write data to an IO object. If offset is -1, write at the object
- maintained file pointer. If the object is not seekable, offset is
- ignored. The amount successfully written is returned in amount. A
- given user should not have more than one outstanding io_write on an
- object at a time; servers implement congestion control by delaying
- responses to io_write. Servers may drop data (returning ENOBUFS)
- if they recevie more than one write when not prepared for it. */
+/* Write data to an IO object. If offset is -1, write at the object maintained
+ file pointer. If the object is not seekable, offset should be -1 or ESPIPE
+ would be returned. The amount successfully written is returned in amount. A
+ given user should not have more than one outstanding io_write on an object
at
+ a time; servers implement congestion control by delaying responses to
+ io_write. Servers may drop data (returning ENOBUFS) if they recevie more
than
+ one write when not prepared for it. */
kern_return_t
trivfs_S_io_write (struct trivfs_protid *cred,
mach_port_t reply, mach_msg_type_name_t replytype,
diff --git a/trans/streamio.c b/trans/streamio.c
index 0faa8a3..07a0d12 100644
--- a/trans/streamio.c
+++ b/trans/streamio.c
@@ -485,6 +485,9 @@ trivfs_S_io_read (struct trivfs_protid *cred,
if (!(cred->po->openmodes & O_READ))
return EBADF;
+ if (offs != (off_t) -1)
+ return ESPIPE;
+
mutex_lock (&global_lock);
err = dev_read (amount, (void **)data, data_len, cred->po->openmodes &
O_NONBLOCK);
mutex_unlock (&global_lock);
@@ -524,6 +527,9 @@ trivfs_S_io_write (struct trivfs_protid *cred,
if (!(cred->po->openmodes & O_WRITE))
return EBADF;
+ if (offs == (off_t) -1)
+ return ESPIPE;
+
mutex_lock (&global_lock);
err = dev_write ((void *)data, data_len, amount, cred->po->openmodes &
O_NONBLOCK);
mutex_unlock (&global_lock);
HTH,
--
Manuel Menal
signature.asc
Description: OpenPGP digital signature
