,
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
In data sabato 28 novembre 2020 20:02:52 CET, Samuel Thibault ha scritto:
> Hello Pino,
>
> Pino Toscano, le sam. 28 nov. 2020 19:45:12 +0100, a ecrit:
> > Can AT_NO_AUTOMOUNT be (easily) implemented in the hurd lookup bits?
>
> I would say that it simply maps to O_NOTRANS?
"return -err" in
make_sockaddr_port() after the getname call, or in all the callers of
make_sockaddr_port().
--
Pino Toscano
#include
#include
#include
#include
#include
#include
int main()
{
int s, e;
struct sockaddr_in ss;
socklen_t len;
s = socket(AF_INET, SOCK_STREAM,
Alle lunedì 11 ottobre 2010, Samuel Thibault ha scritto:
> Pino Toscano, le Sun 10 Oct 2010 23:46:10 +0200, a écrit :
> > make_sockaddr_port() calls the getname function of the sock-ops of
> > the current sock struct, which can be one of
> > - linux-src/net/ipv6/af_
renaming the variable to ED_ (with a
patch in the Debian packaging), otherwise many important packages
couldn't have been compiled.
> Couldn't it be renamed to something less generic?
Or just removed, IMHO.
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
le system ID. */
#define st_dev st_fsid
Is this allowed by POSIX (most probly I'm missing the right part(s) of
it)?
If not, would a "fix" for the above be swapping member and define, like:
__fsid_t st_dev; /* File system ID. */
#define st_fsid st_dev
?
Thanks,
--
P
trying to setsockopt(SO_REUSEADDR) on an AF_UNIX
socket, which makes pflocal return EOPNOTSUPP.
By default, the warning message for this setsockopt(9 is not fatail
though.
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
yntax here)
> +iproute [linux-any]| inetutils-tools [hurd] , ${shlibs:Depends},
>
> ${misc:Depends}
iproute [linux-any] | inetutils-tools [hurd-any]
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
* kern/mach_clock.c (monotonic, mmonotonic): New variables.
(clock_interrupt): Call `time_value_add_usec' also on `monotonic', and
`update_mapped_time' also for `mmonotonic'.
(mapable_time_init): Call `mapable_time_init_time' also `monotonic' and
`mmonotonic'.
---
kern/mach_clock.c |6 ++
Hi,
after Samuel's hint about mapped time, I've been able to add a new clock
in GNU Mach, only updated on clock interrupts.
Also, this clock is exposed like as device like the "time" one.
Pino Toscano (4):
Make update_mapped_time() take also the mapped time variable
Mak
* kern/mach_clock.c (update_mapped_time): Add a second parameter for the
mapped time variable, instead of harcoding it to `mtime'.
---
kern/mach_clock.c | 16
1 files changed, 8 insertions(+), 8 deletions(-)
diff --git a/kern/mach_clock.c b/kern/mach_clock.c
index 4ba7c08..af22
* kern/mach_clock.c (mapable_time_init_time): New function doing the job
previously done by `mapable_time_init', taking the time variables and the
string for the job.
(mapable_time_init): Call `mapable_time_init_time' for `time' and `mtime'.
---
kern/mach_clock.c | 17 +
1 files
* i386/i386at/conf.c (dev_name_list): Add a new `monotonic' device.
* i386/i386at/model_dep.c (monotonicmmap): New function to map `mmonotonic'.
---
i386/i386at/conf.c |8
i386/i386at/model_dep.c | 15 +++
2 files changed, 23 insertions(+), 0 deletions(-)
diff --g
.
Attached there is a git-format'ed patch to fix all (hopefully) of those
cases.
--
Pino Toscano
From 2ee8a64be2d79eb006eebbac95c859f1fc7676f0 Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Tue, 18 Oct 2011 23:46:36 +0200
Subject: [PATCH] Add missing format strings for error, printk, problem
f PATH_MAX .
undefined
--
Pino Toscano
From f7835870d3a2f0447faa31f1159a50a63e26eac9 Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Tue, 18 Oct 2011 23:47:51 +0200
Subject: [PATCH] libdiskfs: handle _PC_PATH_MAX in pathconf
Explicitly return -1 also for _PC_PATH_MAX to indicate there is no
li
e
attached patch should do it (testing welcome!).
--
Pino Toscano
From 2ad3fb8ec4969144536b0e1a877971abae4224c3 Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Wed, 19 Oct 2011 22:25:34 +0200
Subject: [PATCH] libnetfs: handle _PC_PATH_MAX in pathconf
Explicitly return -1 also for _PC_PATH_MAX
/open_issues/libpthread_pthread_key_create_reuse.html
--
Pino Toscano
From 557337b9348f29b29acb4b226fd8ffddf5f22d30 Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Wed, 2 Nov 2011 17:38:46 +0100
Subject: [PATCH] Remove all the values when deleting a key
When deleting a key using
Hi,
attached there is a patch to do validity checks for the pthread key
passed to pthread_getspecific and pthread_setspecific (more details in
the commit log), with two very simple tests for the new checks.
Thanks,
--
Pino Toscano
From e64c1c3f36168593eccb577218971d4fa0448ee0 Mon Sep 17 00:00
ut (when specified) to at least 1ms is usually enough
Note both these two problems also prevent gamin to work, and the
select() issue also may affect other software.
[1] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=79358
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
Alle sabato 27 agosto 2011, Samuel Thibault ha scritto:
> Roland McGrath, le Sat 27 Aug 2011 10:03:24 -0700, a écrit :
> > > Pino Toscano, le Fri 26 Aug 2011 20:29:43 +0200, a écrit :
> > > > +extern vm_offset_t monotonicmmap();
> > > > +#define
example), causing also a warning "task XXX deallocating an invalid
port YYY, most probably a bug" in that situation.
--
Pino Toscano
Refuse negative socket domains right away; otherwise, it is possible to read
and set out-of-bounds locations of the `servers' array (returning the
Alle lunedì 21 novembre 2011, Samuel Thibault ha scritto:
> Pino Toscano, le Mon 21 Nov 2011 01:25:55 +0100, a écrit :
> > attached there is a patch for glibc to check for negative indexes
> > in _hurd_socket_server().
>
> Thanks! Could you submit it to libc-alpha? (with r
Alle lunedì 21 novembre 2011, Samuel Thibault ha scritto:
> Pino Toscano, le Mon 21 Nov 2011 01:25:55 +0100, a écrit :
> > attached there is a patch for glibc to check for negative indexes
> > in _hurd_socket_server().
>
> Thanks! Could you submit it to libc-alpha? (with r
fully cope with this situation?
On Linux the address length is set to 0 by recvfrom(), so I guess that
this kind of sockets have no address?
--
Pino Toscano
#include
#include
#include
#include
#include
#include
#include
void die(int x, const char *s)
{
perror(s);
exit(x);
}
int main()
{
Alle mercoledì 23 novembre 2011, Thomas Schwinge ha scritto:
> Hi!
>
> On Mon, 21 Nov 2011 22:23:26 +0100, Pino Toscano
wrote:
> > Reject negative socket domains right away; otherwise, it is
> > possible to read and set out-of-bounds locations of the `servers'
> >
Hi,
attached there is a patch for the Hurd recvfrom(). It avoids the
application to get a SIGLOST as a indirect consequence of a null mach
port being used.
Thanks,
--
Pino Toscano
hurd: recvfrom(): take into account null address ports
Some kinds of sockets may return a null address port when
to sizeof(addr). */
if (sockaddrlen != sizeof(addr)) {
croak("Bad arg length for %s, length is %d, should be %d",
"Socket::unpack_sockaddr_un",
sockaddrlen, sizeof(addr));
}
# endif
--
Pino Toscano
#in
Alle domenica 4 dicembre 2011, Samuel Thibault ha scritto:
> Pino Toscano, le Sun 04 Dec 2011 13:14:52 +0100, a écrit :
> > while debugging, I apparently stumbled upon a bug of getsockname()
> > for a bound unix socket, as also the attached test program shows:
> > sun_path
Hi,
attached there is a patch to fix in pflocal and pfinet the return values
for wrong values of socket types and protocols on socket creation.
--
Pino Toscano
From 904cbb3dbf1478995629f0cfb83713de786c25e8 Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Tue, 6 Dec 2011 00:30:30 +0100
Hi,
attached there is a patch to add some validation for parameters of
_hurd_select().
Thanks,
--
Pino Toscano
_hurd_select: check for invalid parameter values
Check for invalid values of the `timeout' and `nfds' parameters; move the
calculation of `to' right after the validat
Hi,
attached there is a patch to add some validation for parameters of
Mach's nanosleep() (used by Hurd).
Thanks,
--
Pino Toscano
mach: nanosleep: check for invalid parameter values
Check for invalid values of the `requested_time' parameters; move the
calculation of `ms' after
file.
I didn't have much success in locating who is replying to the io_read
for those fd's, so for now I'll just report the issue.
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
/master branch) to fix this
behaviour in per-process `stat' and `status' files.
Thanks,
--
Pino Toscano
From 3eec2e26d3c4cf83004eed079f0455ed4485095e Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Sat, 14 Jan 2012 14:47:58 +0100
Subject: [PATCH] PID stat/status: show only the fil
Hi,
attached there is a small patch to fix the errno value for the "protocol
not supported" case, from the non-standard EPFNOSUPPORT to EAFNOSUPPORT.
Thanks,
--
Pino Toscano
2012-01-22 Pino Toscano
* sysdeps/mach/hurd/socket.c (__socket): Return EAFNOSUPPORT instead
of the no
nk?
Could it be making our tools like swapon/mount/etc a bit more compatible
(in term of command line options) with the other OSes' ones a (low
priority) goal?
--
Pino Toscano
From d34079adfadebc5f0e936e92257811676b2ebe88 Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Sat, 28 Jan 2012 22
lock.diff:
This patch actually adds the actual new clock, taking care of adding
microseconds to it, and switching the mapped time struct to the new
_extended one.
As usual, comments, remarks, etc are welcome.
(I will provide clean git commits with proper changelog once the work i
finalized.)
T
.
The only bit I'm not too sure is the linking to rt in case libpthread is
compiled a glibc addon (as result of the changes by Samuel an hour ago).
--
Pino Toscano
From d0c2da9c229d90c2301f2263c0b8856ceaf9f517 Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Sun, 22 Apr 2012 00:22:47 +02
of
failing).
Thanks,
--
Pino Toscano
hurd: compliance fixes for getgroups
Fail with EINVAL when the requested number of groups is negative,
or when it is positive but less than the actual number of groups.
2012-04-27 Pino Toscano
* sysdeps/mach/hurd/getgroups.c (__getgroups): Return -1 and set
Hi,
attached there is a patch to fix few issues in Hurd's getlogin_r():
avoid a static buffer, and check for buffer shorter than necessary.
Thanks,
--
Pino Toscano
hurd: compliance fixes for getlogin_r
* do not make `login' static, as it would make getlogin_r no more reentrant
*
Hi,
attached there is a patch to fix few issues in Hurd's ptsname_r(),
mostly checking for more error conditions and making sure to set as
errno and return the proper values on error conditions.
Thanks,
--
Pino Toscano
hurd: compliance fixes for ptsname_r
ptsname_r on failure return
Alle venerdì 27 aprile 2012, Roland McGrath ha scritto:
> > 2012-04-27 Pino Toscano
> >
> > * sysdeps/mach/hurd/getgroups.c (__getgroups): Return -1 and set
> > EINVAL for negative `n' or less than `ngids'.
>
> The norm is to use all caps
sarily catch it
> anyway. If you want that sort of paranoia, use __strnlen (login,
> sizeof login - 1).
>
> If you've already called strlen/strnlen then don't use strncpy,
> just use memcpy with the known length.
Fixed.
Attached there is the updated patch.
--
Pino Tosc
tsname_r(fd, NULL, 0) would be properly failing with EINVAL (and the
man page even explicitly mention that return value, unlike with
basically most of the other functions). Should gnulib do that check only
on Linux, then?
Attached the updated patch, also with the hits of the other reviews.
Thanks
Alle sabato 10 dicembre 2011, Pino Toscano ha scritto:
> attached there is a patch to add some validation for parameters of
> Mach's nanosleep() (used by Hurd).
Disregard this patch -- I'll send a better one with more fixes for
nanosleep.
--
Pino Toscano
signature.asc
Descri
Hi,
attached there is a patch to make Mach's nanosleep() (used by Hurd) a
bit more POSIX compliant:
* check for validity of the `requested_time' parameter
* return EINTR when interrupted
* modify correctly the `remaining' parameter only when interrupted
Thanks,
--
Pino Toscano
m
Hi,
currently, sendto() crashes when the specified addr is NULL; the
attached patch fixes it making sendto() with NULL addr behave as it
would be a plain send().
Thanks,
--
Pino Toscano
Hurd: sendto: do not crash when addr is NULL
Work also when the specified addr is NULL; simplify also the
> structure and port are kept alive around the AF_LOCAL case's calls,
> but that seems OK.
Done.
Updated patch attached.
--
Pino Toscano
Hurd: sendto: do not crash when ADDR is NULL
Create a new create_address_port subroutine to isolate the address port creation
(for both local and
aw other cases of shadowing by internal subroutines, and
assumed it was okay (since inside them you want to refer to the local
"err" instead of the outermost one.
Thanks for the further review, patch attached.
(I would use the non-ChangeLog lines of the patch header as commit log,
would that
Hi,
two simple patches to implement llistxattr and lremovexattr the same way
respectively listxattr and removexattr are implemented, just passing
O_NOLINK. This way all the *xattr functions are implemented, at least to
use the internal _hurd_xattr_*.
Thanks,
--
Pino Toscano
Hurd: provide
F_LINK/AF_LINK
defines be the right solution?
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
attached patch comments them out (instead of removing them, to keep
their placeholders).
Thanks,
--
Pino Toscano
Hurd: comment PF_LINK/AF_LINK defines
Comment out the PF_LINK and AF_LINK defines, since they are usually associated
with struct sockaddr_dl, which is not available on Hurd.
2012-06
Hi,
Alle mercoledì 13 giugno 2012, Pino Toscano ha scritto:
> Alle martedì 12 giugno 2012, Roland McGrath ha scritto:
> > That's almost right, but you should avoid calling
> > __mach_port_deallocate if APORT is MACH_PORT_NULL. Also, avoid
> > shadowing the local variabl
Hi,
the attached patch implements renameat for Hurd; it seems to work fine,
and tst-renameat now passes.
The copyright starts from the year of the rename implementation, as I
basically copied it and slightly adapted to the "at" behaviour.
Thanks,
--
Pino Toscano
Hurd: implemen
second.
>
> Otherwise that looks fine, so go ahead and commit it with that
> change.
Ok, I will do.
Thanks,
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
a1_native and
a2_native).
Does it look good?
--
Pino Toscano
Provide a generic empty version of __check_native.
Add an empty implementation of __check_native, as used in the posix version of getaddrinfo.
This allows non-Linux GNU-based OSes to compile.
2012-07-26 Pino Toscano
* inet
POSIX.
Thanks,
--
Pino Toscano
Hurd: fix fdatasync/fsync if the fd does not support file_sync
Handle the case of the fd port not implementing file_sync (returning MIG_BAD_ID) or implementing a stub (EOPNOTSUPP),
properly returning EINVAL.
2012-08-29 Pino Toscano
* sysdeps/mach/hurd/fdatasync.c: Tur
Hi,
attached there is a minor patch to fix the return values of
unimplemented cases, i.e. using __hurd_fail to set EOPNOTSUPP properly.
Thanks,
--
Pino Toscano
Hurd: ptrace: use __hurd_fail for EOPNOTSUPP
2012-08-29 Pino Toscano
* sysdeps/mach/hurd/ptrace.c (ptrace): Use __hurd_fail to
Hi,
attached there is a new patch for procfs to improve again the file name
in stat/status files for PIDs.
Basically, it just considers only the first word in case the process
name has more (e.g. when it changed its own).
Thanks,
--
Pino Toscano
From ebf2b049ea6963026766763df1697467f5806327
,
--
Pino Toscano
From a35bd267aba2954df85ee3403981395bf8c9e736 Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Fri, 7 Sep 2012 18:21:00 +0200
Subject: [PATCH 1/2] tmpfs: extract size string parsing in an own function
* tmpfs/tmpfs.c (parse_opt_size): New function, broken out of ...
(parse_opt
Alle domenica 9 settembre 2012, Samuel Thibault ha scritto:
> Pino Toscano, le Fri 07 Sep 2012 20:02:56 +0200, a écrit :
> > +static int args_filename_length (const char *name)
> > +{
> > + const char *p = name;
> > + while (*p != '\0' && *p != &#x
Alle lunedì 10 settembre 2012, Samuel Thibault ha scritto:
> Pino Toscano, le Mon 10 Sep 2012 17:31:32 +0200, a écrit :
> > Alle domenica 9 settembre 2012, Samuel Thibault ha scritto:
> > > Pino Toscano, le Fri 07 Sep 2012 20:02:56 +0200, a écrit :
> > > > +static i
Hi,
it seems that in some occasions (i.e on malloc failure) pthread_create
can return ENOMEM, which is not a valid POSIX error number.
Attached there is a patch for libpthread to normalize ENOMEM to EAGAIN.
Thanks,
--
Pino Toscano
From 2ee7e1e143784452ef325a0c80c106e972a3ffdc Mon Sep 17 00:00
places.
Otherwise, I'm not yet sure how fix the lookup glibc functions.
As I said, I'm pretty new to the involved parts (lookup stuff), so if
you have more ideas I will hear them.
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
> (http://lists.gnu.org/archive/html/bug-hurd/2012-03/msg00013.html)
> and [made console-run resilient against missing /dev/console]
> (http://lists.gnu.org/archive/html/bug-hurd/2012-03/msg2.html),
> improving the overall reliability of the system.
Not sure how they "improve
le of months.)
Thanks,
--
Pino Toscano
From 399f9112f763c095abe44567135307c554746e9e Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Sun, 23 Sep 2012 23:53:00 +0200
Subject: [PATCH] configure: add --with-version-suffix=STRING
Add the possibility to append a custom string to the gnumach version s
Hi,
Alle lunedì 24 settembre 2012, Thomas Schwinge ha scritto:
> Generally OK, but two comments:
>
> On Mon, 24 Sep 2012 00:04:20 +0200, Pino Toscano
wrote:
> > Subject: [PATCH] configure: add --with-version-suffix=STRING
> >
> > Add the possibility to append a c
umber provided.
Thanks,
--
Pino Toscano
Hurd: mknodat: create also sockets
2012-09-28 Pino Toscano
* sysdeps/mach/hurd/xmknodat.c: Allow to create sockets. Simplify the
check for translators needing the device number.
--- a/sysdeps/mach/hurd/xmknodat.c
+++ b/sysdeps/mach/hurd/xmknodat.c
@@
ince all it does is testing the creation of a single socket.
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
gle socket.
>
> I think the intent of the change is to test the mknodat interface,
> not any particular kind of node. Just change the test to create a
> FIFO instead.
Right, patch attached.
--
Pino Toscano
tst-mknodat: create a FIFO instead of a socket
A FIFO is the only special fi
Alle venerdì 28 settembre 2012, Roland McGrath ha scritto:
> That change looks fine if the test still passes on Linux.
It does on Linux/x86_64, so I pushed it.
Thanks for the reviews,
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
Hi,
attached there is a simple patch to make
(f)pathconf(path, _PC_2_SYMLINKS) working for libdiskfs-based
translators, properly indicating whether the FS supports symlinks.
Thanks,
--
Pino Toscano
From edbb0cc20709e0599e1490b81aa1841114f3f0bc Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date
Hi,
attached there is a patch to add a custom version (copied from the unix
sysdep) of confstr.h, to avoid using the bsd header which adds a non-
existing /usr/ucb.
Thanks,
--
Pino Toscano
Hurd: fix the _CS_PATH getconf value
Copy sysdeps/unix/confstr.h to avoid using sysdeps/unix/bsd
Hi,
The following two patches allow to
a) make use of the "pthread" glibc sysdep, which provides implementation
for some functions using pthreads (e.g. aio stuff)
b) hook the libpthread version inside glibc, so it is properly returned
by confstr.
--
Pino To
Hi,
Alle giovedì 27 settembre 2012, Thomas Schwinge ha scritto:
> On Mon, 24 Sep 2012 02:04:05 +0200, Guillem Jover
wrote:
> > On Mon, 2012-09-24 at 00:30:39 +0200, Pino Toscano wrote:
> > > Alle lunedì 24 settembre 2012, Thomas Schwinge ha scritto:
> > > > On M
Alle martedì 2 ottobre 2012, Thomas Schwinge ha scritto:
> Hi!
>
> On Mon, 1 Oct 2012 19:59:35 +0200, Pino Toscano
wrote:
> > Alle giovedì 27 settembre 2012, Thomas Schwinge ha scritto:
> > > On Mon, 24 Sep 2012 02:04:05 +0200, Guillem Jover
> >
> > wrote:
Hi,
Alle lunedì 1 ottobre 2012, Andreas Jaeger ha scritto:
> On Sunday, September 30, 2012 17:53:33 Pino Toscano wrote:
> > attached there is a patch to add a custom version (copied from the
> > unix sysdep) of confstr.h, to avoid using the bsd header which adds
> > a n
Alle mercoledì 3 ottobre 2012, Thomas Schwinge ha scritto:
> Hi!
>
> On Wed, 3 Oct 2012 17:50:38 +0200, Pino Toscano
wrote:
> > Alle martedì 2 ottobre 2012, Thomas Schwinge ha scritto:
> > > On Mon, 1 Oct 2012 19:59:35 +0200, Pino Toscano
> >
> > wrote:
&g
nly fatfs,
IIRC) do not support symlinks.
I've reverted the commit, and going to propose a better (I hope)
solution for this.
Thanks,
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
binary compatibility,
and a filesystem providing such diskfs_pathconf implementation should
run fine even with an older libdisks (but I did not test this case).
What do you think about this solution?
--
Pino Toscano
--- a/libdiskfs/diskfs.h
+++ b/libdiskfs/diskfs.h
@@ -553,6 +553,9
Alle mercoledì 29 agosto 2012, Pino Toscano ha scritto:
> Hurd's implementations of fdatasync and fsync do not take into
> account the fact that file_sync could not be implemented in the
> receiving port, or implemented as stub, returning (E)MIG_BAD_ID or
> EOPNOTSUPP. Attached th
Alle mercoledì 29 agosto 2012, Pino Toscano ha scritto:
> attached there is a minor patch to fix the return values of
> unimplemented cases, i.e. using __hurd_fail to set EOPNOTSUPP
> properly.
Ping.
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
e (i.e. pflocal), libtrivfs is used, EOPNOTSUPP is
> returned, right?
The run above would have fsync() set errno = -303, MIG_BAD_ID.
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
SIX.1 specifies EINVAL for a file descriptor on which
> sync is not supported.
Attached an updated patch to handle only EOPNOTSUPP, with comment added.
Thanks,
--
Pino Toscano
Hurd: fix fdatasync/fsync if the fd does not support file_sync
Handle the case of the fd port implementing a stub
with glibc < 2.16.
Also, I'm slightly in doubt about the ChangeLog snippet, not totally
sure about how to write it properly in this case.
--
Pino Toscano
From 64d00b6482f03f7dfb6f3d4775a9bf44e53cec63 Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Fri, 16 Nov 2012 15:04:29 +0100
Subjec
Hi,
the attached patch makes check-local-headers.sh ignore headers from the
"mach" subdirectory, since they the Mach kernel headers used to build
glibc on Hurd.
Thanks,
--
Pino Toscano
check-local-headers: ignore Mach kernel headers
2012-11-16 Pino Toscano
* scripts/c
y after that close the file.
Thanks,
--
Pino Toscano
#0 0x0106281c in swtch_pri () at /build/buildd-eglibc_2.13-36-hurd-i386-IvO_gk/eglibc-2.13/build-tree/hurd-i386-libc/mach/swtch_pri.S:2
#1 0x010640a4 in __spin_lock_solid (lock=0x11ded74) at spin-solid.c:27
#2 0x0106429d in __mutex_lock_sol
this be considered
acceptable?
Thanks,
--
Pino Toscano
Simplify ulimit implementations
The ulimit implementation in sysdeps/unix/bsd/ulimit.c produces wrong return
values, while sysdeps/unix/sysv/linux/ulimit.c is generally better. Thus,
copy the latter into a more general sysdeps/unix/ulimit.c
Hi,
(bug-hurd only for this)
Alle sabato 17 novembre 2012, Pino Toscano ha scritto:
> Ignoring the fact that in libio mmap is not used on Hurd (it will
> need a different fix),
This has been reported by Thomas Schwinge one year ago:
http://thread.gmane.org/87mx
Hi,
simple implementation of the Linux-ish syncfs on Hurd.
Thanks,
--
Pino Toscano
Hurd: implement syncfs
2012-11-19 Pino Toscano
* sysdeps/mach/hurd/syncfs.c: New file.
--- /dev/null
+++ b/sysdeps/mach/hurd/syncfs.c
@@ -0,0 +1,33 @@
+/* Copyright (C) 2012 Free Software Foundation, Inc
t; Third commit: remove sysdeps/unix/bsd/ file.
OK, I will split the commits once approved.
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
Hi,
Alle lunedì 19 novembre 2012, Pino Toscano ha scritto:
> simple implementation of the Linux-ish syncfs on Hurd.
Updated patch according to Roland's and Christoph's comments (thanks!)
--
Pino Toscano
Hurd: implement syncfs
2012-11-19 Pino Toscano
* sysdeps/mach/hurd/syncf
of comment added.
--
Pino Toscano
muntrace: reset file and hooks before finalizing the stream
fclose will call free, invoking its hook, then fprintf which would indirectly
try to allocate a buffer, and this can cause malloc to be used (thus its hook
to be invoked) if libio uses malloc instead o
ven the history, the least of
> avaialble evils is to make the Hurd implementation consistent with
> the others and do the check.
(few months later... I forgot I sent this patch, so I'm bring it again.)
I updated the patch; is it okay to commit, or should I bring back the
buf==NULL check?
--
Pi
ig and
> libx11-dev (neither was installed by apt-get build-dep hurd, or
> build-essential or devscripts).
Those have been added by Samuel in the packaging repository some time
ago, and will be part of the next upload.
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
Hi,
attached there is a small patch for procfs to implement statfs,
returning only the two fields we can fill for sure.
May I push it?
Thanks,
--
Pino Toscano
From 1b7ad5c5d601b6388f0fc871b4fa42d231c05400 Mon Sep 17 00:00:00 2001
From: Pino Toscano
Date: Thu, 6 Dec 2012 17:51:58 +0100
Subject
Alle giovedì 6 dicembre 2012, Thomas Schwinge ha scritto:
> Hi!
>
> On Thu, 6 Dec 2012 17:57:32 +0100, Pino Toscano
wrote:
> > Initial implementation of statfs reply, just returning the
> > filesystem type and its id.
> >
> > * netfs.c: Include and .
>
, the case for having it at all is
> inordinately weak.
OK.
Attached there are the two patches of the ulimit reorganization, as you
suggested in a previous email:
1) move the linux implementation as posix (including limits.h)
2) remove the bsd implementation
Thanks,
--
Pino Toscano
ulimit: move lin
d[i].ulink); if (d[i].io_port != MACH_PORT_NULL)
> continue;
> - }
why is this check removed? When collecting inputs from poll,
d[i].io_port is the fd passed, which has no guarantee to be lower than
_hurd_dtablesize.
--
Pino Toscano
signature.asc
Description: This is a digitally signed message part.
Alle martedì 22 gennaio 2013, Svante Signell ha scritto:
> On Tue, 2013-01-22 at 19:15 +0100, Pino Toscano wrote:
> > Alle martedì 22 gennaio 2013, Svante Signell ha scritto:
> > > Attached is the first patch for a 3-way split of hurdselect.c
> > > into three
ally, do a commit removing the bsd file.
Done, I think I should have done everything correctly.
The addition of was needed on Hurd (I tested it before
sending the first version of this ulimit rework) -- in any case,
LONG_MAX is explicitly used, so that include is not wrong.
Thanks,
--
Pino Toscan
1 - 100 of 135 matches
Mail list logo
