* ioctl-tests/qioctl-handler.c (hurd_ioctl_handler):
Update to reflect `ioctl_handler_t' change.
---
ioctl-tests/qioctl-handler.c | 20 +++-
1 files changed, 11 insertions(+), 9 deletions(-)
diff --git a/ioctl-tests/qioctl-handler.c b/ioctl-tests/qioctl-handler.c
index a5b2503..
* (ioctl-tests): New subdirectory.
* (ioctl-tests/Makefile)
(ioctl-tests/qio.h)
(ioctl-tests/qioctl.c)
(ioctl-tests/qioctl.defs)
(ioctl-tests/qioctl-handler.c)
(ioctl-tests/test.c): New files.
---
Makefile |2 +-
ioctl-tests/Makefile | 45 +
ioctl-test
* Makefile (ioctl_handler_MIGSFLAGS): New variable.
* ioctl-tests/qioctl.c (S_ioctl_handler_get): Remove deprecated routine.
(S_ioctl_handler_request): New funtction.
---
ioctl-tests/Makefile |1 +
ioctl-tests/qioctl.c | 29 -
2 files changed, 25 insertions(+), 5
Hi,
Here comes patches that provide tests for server provided
ioctl-handlers. These should be applied in parallel to the
glibc patches, to which they correspond one-to-one.
Since there aren't many test cases in the Hurd, I didn't have much
to go by in implementing the test. Instead I just focus
* hurd/Makefile (interfaces): Add `ioctl_handler_reply'.
* hurd/fd-ioctl-call.c: Check that handlers are provided by the same user.
---
hurd/Makefile|3 +-
hurd/fd-ioctl-call.c | 157 +-
2 files changed, 158 insertions(+), 2 deletions(-)
* hurd/hurd/ioctl.h (ioctl_handler_t): Move from here...
* hurd/hurd/fd.h (ioctl_handler_t): ...to here.
Change return type.
Add `d', `crit', and `result' arguments.
Change all callers and ioctl handlers.
(hurd_fd): Add ioctl handler members.
* hurd/fd-close.c: Update copyright years.
(_hurd_fd_clo
* hurd/hurdioctl.c (_hurd_dummy_ioctl_handler): New function.
* hurd/hurd/ioctl.h (_hurd_dummy_ioctl_handler): Likewise.
* hurd/fd-ioctl-call.c: New file.
* hurd/hurd/fd.h: Update copyright years.
(_hurd_fd_call_ioctl_handler): New function declaration.
* hurd/Makefile: Update copyright years.
(use
Hi,
Patches that makes glibc make use of the new ioctl_handler protocol.
This should be applied in parallel to the protocol patches; the first
two patches depend on the first protocol patch, the third one depend on
the second protocol patch. In addition all these patches depend on the
three clean
* hurd/ioctl_handler.defs (ioctl_handler_get): Remove routine.
(ioctl_handler_request): New routine.
(ioctl_handler_reply): Allocate space for this routine.
* hurd/ioctl_handler_reply.defs: New file.
---
hurd/ioctl_handler.defs | 37 +++-
hurd/ioctl_handler_repl
* hurd/ioctl_handler.defs: New file.
* hurd/subsystems: Add ioctl_handler.
---
hurd/ioctl_handler.defs | 35 +++
hurd/subsystems |1 +
2 files changed, 36 insertions(+), 0 deletions(-)
create mode 100644 hurd/ioctl_handler.defs
diff --git a/hurd/ioct
Hi,
this patch series adds a MIG interface for getting an ioctl handler module
associated with an io object. The first patch adds a plain and insecure
protocol, while the second does reverse authentication to establish the
identity of the module provider.
Regards,
Fredrik
Carl Fredrik Hammar
* hurd/lookup-retry.c (__hurd_file_name_lookup_retry) :
Reopen file descriptor before returning it.
---
Hi,
Another stand-alone patch, but this one is new. It fixes what I think is
a bug in glibc.
The problem is that when opening file descriptors using FS_MAGIC_RETRY's
`fd/*' syntax, the descrip
* hurd/hurdioctl.c (tiocsctty): Only get FD ports, do work in...
(tiocsctty_internal): ...this new function.
---
Hi,
This is another stand-alone patch I have sent earlier.
This fixes the handler for TIOCSCTTY so it only resolves the underlying
port of the file descriptor once. Since the descript
* hurd/hurd/ioctl.h (_HURD_HANDLE_IOCTLS_1): Cast to `ioctl_handler_t'.
---
Hi,
This is an obvious fix to glibc that makes it easier to change
`ioctl_handler_t'. I sent in this patch before as part of the server
provided ioctl handler patch series. I'm resending it because I'm going
to resend th
14 matches
Mail list logo