name of the paper.
It disappeared a few months ago, and since it's not widespread, I
suggest, despite providing to Justus in the first place, to use the
Murmur finalizer instead. But we need to test that part a bit more.
I'm currently doing just that.
--
Richard Braun
nts_t, I like the idea of using the
> > type system to enforce the use of the accessor functions.
>
> I understand, but type-puning will break with smart compilers. We do
> have had bugs like that in the past.
I agree, you should directly use the proper type here, or maybe void *
but I don't see the point.
--
Richard Braun
On Thu, Feb 27, 2014 at 10:17:07PM +0100, Richard Braun wrote:
> I intend to regularly update these packages to track the experimental
> branch until the changes are merged in the official repository.
Iceweasel 29.0.1 (from unstable) is available.
--
Richard Braun
sic infrastructure for the trap call.
Keep in mind that, while most "system calls" are actually RPCs, a few of
them actually are real system calls (i.e. traps) in addition to
mach_msg. mach_thread_self and mach_task_self are such calls.
--
Richard Braun
On Tue, May 20, 2014 at 11:50:28PM +0200, Samuel Thibault wrote:
> Richard Braun, le Tue 20 May 2014 23:42:24 +0200, a écrit :
> > On Tue, May 20, 2014 at 10:50:07PM +0200, Samuel Thibault wrote:
> > > I would advise starting with trivial system calls, such as
>
On Tue, May 13, 2014 at 10:37:05AM +0200, Richard Braun wrote:
> It disappeared a few months ago, and since it's not widespread, I
> suggest, despite providing to Justus in the first place, to use the
> Murmur finalizer instead. But we need to test that part a bit more.
> I'
On Fri, May 23, 2014 at 01:44:09AM +0200, Samuel Thibault wrote:
> Richard Braun, le Mon 05 May 2014 18:32:26 +0200, a écrit :
> > On Mon, May 05, 2014 at 06:01:17PM +0200, Samuel Thibault wrote:
> > > ? The patch makes both ext2fs's service_paging
t; time, and thus not pipelining the I/O requests.
Right, I assumed the store interface wasn't synchronous...
--
Richard Braun
On Fri, May 30, 2014 at 09:35:37AM +0200, Justus Winter wrote:
> Quoting Richard Braun (2014-05-29 19:12:13)
> > On Thu, May 29, 2014 at 07:04:48PM +0200, Samuel Thibault wrote:
> > > But precisely: once the only thread gets a data_request, it'll call
> > > pager
On Fri, May 30, 2014 at 08:15:43PM +0200, Samuel Thibault wrote:
> Justus Winter, le Fri 30 May 2014 09:35:37 +0200, a écrit :
> > Quoting Richard Braun (2014-05-29 19:12:13)
> > > On Thu, May 29, 2014 at 07:04:48PM +0200, Samuel Thibault wrote:
> > > > But precisel
the Hurd is part
> of why I advocated this merge in the first place (as clearly stated in
> my original mail).
I strongly agree with Justus on this.
--
Richard Braun
On Thu, Feb 27, 2014 at 10:17:07PM +0100, Richard Braun wrote:
> I intend to regularly update these packages to track the experimental
> branch until the changes are merged in the official repository.
Iceweasel 30.0-2 (from unstable) is available.
--
Richard Braun
rs to a socket and the flags argument is 0,
the send() function is equivalent to write()."
And here is write() :
"If nbyte is zero and the file is not a regular file, the results are
unspecified."
We might also want to change this though, since the behaviour observed
on other systems seems more appropriate.
--
Richard Braun
On Sat, Jun 28, 2014 at 10:48:56AM +0200, Richard Braun wrote:
> On Sat, Jun 21, 2014 at 03:56:46PM +0200, Andreas Cadhalpun wrote:
> > This is because the client is calling:
> > send(sockfd, "", 0, 0)
> >
> > Normally this doesn't trigger recv() in t
On Sat, Jun 28, 2014 at 12:09:15PM +0200, Samuel Thibault wrote:
> Richard Braun, le Sat 28 Jun 2014 11:51:42 +0200, a écrit :
> > On Sat, Jun 28, 2014 at 10:48:56AM +0200, Richard Braun wrote:
> > > On Sat, Jun 21, 2014 at 03:56:46PM +0200, Andreas Cadhalpun wrote:
> >
On Sat, Jun 28, 2014 at 12:42:40PM +0200, Richard Braun wrote:
> I'll see if simply catching completely empty messages at socket_send is
> a good enough solution.
The solution seems to work, and I couldn't see anything against it,
unlike the previous attempt. However I'd
, but we probably want to align on what others do, as usual.
--
Richard Braun
On Thu, Jul 10, 2014 at 11:18:17PM +0200, Richard Braun wrote:
> All right, it looks like open() gets interrupted by SIGCHLD here. It's
> my understanding that signal handling is highly system-specific in such
> cases, but we probably want to align on what others do, as usual.
On Thu, Jul 10, 2014 at 02:21:06PM -0700, Roland McGrath wrote:
> Is the SIGCHLD set with SA_RESTART?
>From what I see, no.
--
Richard Braun
the returned mapping must not start at
address 0).
--
Richard Braun
On Thu, Feb 27, 2014 at 10:17:07PM +0100, Richard Braun wrote:
> I intend to regularly update these packages to track the experimental
> branch until the changes are merged in the official repository.
Iceweasel 31.1.0esr-1 (from unstable) is available.
--
Richard Braun
on't try
and boss other people. Mike will handle the bug when he decides to, and
I'll keep providing packages in the mean time.
--
Richard Braun
de specific points in time that others can base their work on
(the Nix-based distribution comes to mind).
--
Richard Braun
ssue I
have when building "large" files (see [1]).
--
Richard Braun
[1] https://www.sceen.net/iceweasel-31-on-debian-gnuhurd/
ontinuing.
> ^C
> [same]
Same here, the pageout daemon does use miliseconds sleeps until its load
decreases, which I assume never does in this case.
--
Richard Braun
on doesn't directly
concern Debian.
As for the question itself, the Hurd runs on i386 compatible hardware,
so Intel (including Pentium) and AMD processors are supported.
--
Richard Braun
orward to seeing this in the Debian packages :).
--
Richard Braun
andles the introspection port
> registration and lookup, and the message wrapping/unwrapping.
Well, one big step towards lsof-like tools and easier debugging.
As usual, very good job :).
--
Richard Braun
n see a timeout at the client side. Why a timeout
instead of letting the user interrupt the process, or automated tools
use a higher-level monitoring scheme ?
--
Richard Braun
an entirely different
problem. As long as tools provide an option to tune/disable it, I
don't mind.
--
Richard Braun
p you posted.
It looks like a netdde-specific problem. Perhaps a small mistake with
regard to the static library it includes. Other than that, the patch
looks good, and I'm very, very happy to finally see something like this
land in libpager.
--
Richard Braun
p you posted.
Stressing the file system (by building a large package such as
iceweasel) triggers a reference counting assertion [1]. We'll have to
squash this bug hard before the patch can be merged.
--
Richard Braun
[1]
https://www.gnu.org/software
On Tue, Oct 28, 2014 at 05:59:16PM +0100, Richard Braun wrote:
> Stressing the file system (by building a large package such as
> iceweasel) triggers a reference counting assertion [1]. We'll have to
> squash this bug hard before the patch can be merged.
Actually, after applyin
On Thu, Oct 30, 2014 at 04:43:53PM +0100, Richard Braun wrote:
> Actually, after applying the libpager patch only, as opposed to using
> the continuous integration packages which merge all of Justus' work,
> things look much better. So far I haven't been able to reproduce th
port.
> thank you.
You'll probably end up reinstalling again, just to be safe.
Can we have details about the hardware, such as :
- processor
- memory
- disk interface (IDE/SATA)
- network board
Sorry to ask again if you already provided those.
--
Richard Braun
being actually done for that. From what I
remember, ioctl-specific calls must be added into glibc.
--
Richard Braun
On Thu, Nov 06, 2014 at 04:33:26PM +0100, Samuel Thibault wrote:
> In my memory, it's something like migrating threads which could make it
> possible for more flexible ioctls.
Not migrating threads, object mobility [1].
--
Richard Braun
[1] http://users.student.lth.se/cs07fh9/2009-
then. Other opinions?
It looks good to me too.
--
Richard Braun
On Wed, Nov 12, 2014 at 04:49:07PM +0100, Justus Winter wrote:
> Overall it's looking good, time to get the discussion going.
What privilege is required to request these notifications ?
--
Richard Braun
On Thu, Nov 13, 2014 at 12:09:03PM +0100, Justus Winter wrote:
> Possession of the privileged host control port, and it is only
> possible to register for these notifications once.
How does this recurse in the subhurd ?
--
Richard Braun
On Thu, Nov 13, 2014 at 12:44:12PM +0100, Justus Winter wrote:
> Well, the proc server registers for these notifications. I'll add
> this RPC to the process protocol:
This looks good to me. I like that it relies on the kernel for
security, but that it's also minimalist.
--
Richard Braun
velopment release
> with maximum 200 Kib uncompiled size.
> As less assembler as possible would be great.
Unless you really insist on running the Hurd, I suggest you look at one
of the L4 variants. Mach and the Hurd have always consumed much more
than that.
--
Richard Braun
e?
Yes, please avoid alloca unless it really makes sense.
--
Richard Braun
is
equivalent to acquiring/releasing a big kernel lock. If the critical
sections protected by this lock are all small enough, granularity
tends to matter much less. My point being that this could be generalized
for all IPLs, since they are all affected by the virtualization penalty.
--
Richard Braun
On Sun, Dec 14, 2014 at 03:44:44PM +0100, Richard Braun wrote:
> No, and in fact, I don't like the idea of SPL/IPL. For example, Linux
> doesn't care about them and merely disables or enables interrupts at
> the processor level, the semantics being all or nothing, and it sho
On Sat, Dec 20, 2014 at 11:40:05PM +0100, Samuel Thibault wrote:
> hwloc-bind pu:0 qemu foobar
taskset also does the job.
--
Richard Braun
On Fri, Dec 26, 2014 at 01:58:28PM +0100, Justus Winter wrote:
> It requires only tiny tweaks to libdiskfs, and a tiny patch to gnumach
> that enables us to load non-elf files into tasks.
What kind of non-elf files ?
--
Richard Braun
On Fri, Dec 26, 2014 at 06:32:33PM +0100, Justus Winter wrote:
> Quoting Richard Braun (2014-12-26 17:10:19)
> > What kind of non-elf files ?
>
> Any file. I created this patch in an attempt to use a dead task
> (i.e. a task w/o a thread) as a ramdisk. Now I'm (ab)using t
On Thu, Jan 22, 2015 at 10:22:04PM +0100, Justus Winter wrote:
> So do we consider {deallocating,destroying,mod_refing} MACH_PORT_NULL
> or MACH_PORT_DEAD an error, or an expensive nop ?
Personally, I'd consider this a nop, like free(NULL) is.
--
Richard Braun
> data any more. AIUI this is the same kind of usage that you'd have,
> what do you think?
Personally, I like this better.
--
Richard Braun
7;s odd. Though I must admit I do not use passive translator
> records to mount disks. Maybe the Debian init scripts just aren't
> compatible with that. Personally I'd be ok with that.
Don't mix fstab with passive translators. The init scripts actually
do mount fstab entries now.
--
Richard Braun
emory managment as well).
--
Richard Braun
anagement enough to answer it.
I'd say that glibc (in mmap) should add a reference to the file. On
munmap or process termination, all references are dropped.
--
Richard Braun
On Sat, Feb 28, 2015 at 01:05:07PM +0100, Richard Braun wrote:
> I'd say that glibc (in mmap) should add a reference to the file. On
> munmap or process termination, all references are dropped.
Forget this, it obviously can't work since mmap and munmap can be
asymmetric.
--
Richard Braun
On Mon, Mar 02, 2015 at 05:09:04PM -0500, A. Fleckenstein wrote:
> In the hurd sources, hurd/hurd/fs.defs, line 59 begins a comment, the
> contents of which are an excerpt from the prologue of Geoffrey Chaucer's
> Canterbury Tales. Anyone know why this is?
"chauthor"
--
Richard Braun
On Thu, Feb 27, 2014 at 10:17:07PM +0100, Richard Braun wrote:
> I intend to regularly update these packages to track the experimental
> branch until the changes are merged in the official repository.
Iceweasel 36.0-2 (from experimental) is available.
--
Richard Braun
On Wed, Mar 18, 2015 at 01:52:22PM -0400, A. Fleckenstein wrote:
> Debayan Das writes:
>
> Perhaps someone else can confirm this, but I believe that GNU/Hurd is
> not a mentoring organization this year.
We are [1], and it would be nice to be listed on the GSoC page.
--
Richard Bra
ommunity/gsoc/project_ideas.html
2/ Read
http://www.google-melange.com/gsoc/document/show/gsoc_program/google/gsoc2015/help_page
3/ Also read
http://www.di.ens.fr/~baghdadi/TXT_blog/5_advices_to_get_your_proposal_accepted.lyx.html
Then submit your proposal.
Good luck.
--
Richard Braun
On Fri, May 01, 2015 at 01:00:33PM +0200, Justus Winter wrote:
> What do you think?
That's what I had in mind too.
--
Richard Braun
/
deb-src http://ftp.sceen.net/debian-hurd-i386 experimental/
This code is currently being polished and should be merged soon into
the main gnumach repository.
--
Richard Braun
[1]
https://www.google-melange.com/gsoc/project/details/google/gsoc2015/teythoon/588030830336
On Tue, May 19, 2015 at 05:38:58PM +0200, Justus Winter wrote:
> I have polished the patch series and all commit messages. Please have
> a look and poke me if anything stands out. I've also pushed the
> changes to darnassus:
This looks all good to me, feel free to merge.
--
Richard Braun
a leak. This was the main
difficulty with thread destruction in libpthread. I don't remember
if there were leaks elsewhere.
--
Richard Braun
n ?".
Despite diskfs_remount calling ports_inhibit_class_rpcs, other threads
can very well be running to process previously received messages. There
seems to be no other form of access synchronization such as locks in
diskfs_reload_global_state.
Can you get the call trace leading to ext2_getblk
On Mon, Jul 06, 2015 at 02:32:42PM +0200, Samuel Thibault wrote:
> I got a rdxtree-related assertion error, here is a snapshot.
Well, that's unfortunate :-/. Node insertion is quite a low level
function, we'd need a more complete trace to get a direction where
to look.
--
Richard Braun
Can you manage to get -O0 traces ?
--
Richard Braun
eck that stored pointers aren't NULL and are
correctly aligned. Clearing an entry (rdxtree_node_remove) does set
an entry to NULL, so the LSB should get back to 0 in all cases.
I really don't see how an entry could be 1...
--
Richard Braun
and it was decided we would put our
own RPCs in gnumach.defs. The mach4.defs file is really meant to contain
stuff that came with Mach 4.
--
Richard Braun
cfs --usage' for more information.
mount: cannot start translator /hurd/procfs: Translator died
2/ find: `/var/cache/pbuilder/build//1711/servers/startup': (ipc/mig) bad
request message ID
I guess it's just a matter of updating a few command lines.
--
Richard Braun
eds to be
> paged in when remounting?
Remounting can require paging out, yes.
See diskfs_reload_global_state in ext2fs :
diskfs_reload_global_state ()
{
pokel_flush (&global_pokel);
pager_flush (diskfs_disk_pager, 1);
...
--
Richard Braun
On Tue, Jul 07, 2015 at 04:57:27PM +0200, Richard Braun wrote:
> Even with a clean trace, I have a hard time understanding how it could
> happen.
Unfortunately, rdxtrees are currently unsafe to use without general
(sleep) locks, since they can allocate memory in the middle of an
operation.
strongly advised to check the integrity of their files.
Concerning the bug itself, I suspect it was caused by an LBA28/LBA48
issue, because it happened right after resizing the file systems on
a disk which then got larger than 128G.
--
Richard Braun
it running on
busy buildds.
--
Richard Braun
based
critical sections ?
--
Richard Braun
On Fri, Jul 17, 2015 at 01:13:30AM +0200, Richard Braun wrote:
> Those patches look good but since we don't have a good reproducer for
> this particular problem, the only way to know is to get it running on
> busy buildds.
Actually, we do have a good reproducer.
I inserted ca
rwlock-based synrchonization
functions specific to the pager workers.
--
Richard Braun
adly.
>
> What are your thoughts on that?
I'm completely opposed to this approach. The Hurd has no other purpose
than showing what its elegant and powerful way can provide. I for one
really don't care about getting a user base, unless that user base
exists precisely because of the features of the system.
--
Richard Braun
b-src http://ftp.sceen.net/debian-hurd-i386 experimental/
The only negative consequence I can see is that URLs that relied on
URL rewriting are now invalid, since it's not implemented by this
web server.
--
Richard Braun
[1] http://darnassus.sceen.net/gitweb/
est then, let alone
> Hurd developers (they use kvm/qemu), so issues with Virtual Box are most
> often not even known about.
Most notably, VirtualBox has clock issues, which are very annoying for
practical usage.
--
Richard Braun
xactly, perhaps 15m,
> > certainly less than an hour.
>
> Confirmed to be 12m, from booting the CD to getting an installed system
> prompt.
Keep in mind QEMU provides writeback caching by default now.
--
Richard Braun
ore physical memory and
very strongly reduce kernel memory exhaustion panics.
--
Richard Braun
[1] http://darnassus.sceen.net/gitweb/?p=rbraun/gnumach.git;a=summary
virtual machines.
--
Richard Braun
- @val = compute_values($hz);
> > > - }
> > > + @val = compute_values($hz);
> >
> > Don't we want to still try to use canned_values?
>
> As I wrote in IRC you should solve this, not me. Obviously the !defined(@val)
> is
Apparently you'r
On Mon, Jan 11, 2016 at 04:56:31PM +0100, Thomas Schwinge wrote:
> Hi!
>
> On Sun, 22 Nov 2015 01:01:09 +0100, Richard Braun wrote:
> > It looks like some changes in the past months have increased the
> > severity of some bugs apache is having when spawning CGIs, result
gt; output to be explicitly redirected by callers of mount and settrans?
> In the mount case, people used to Linux will not think of doing it, so
> perhaps we need it to do it for the user?
I think we could have a default log server for that. We could make it
a gsoc project. It should make a good entry project.
--
Richard Braun
ernel.
--
Richard Braun
[1] https://lists.gnu.org/archive/html/bug-hurd/2012-07/msg00031.html
[2]
http://darnassus.sceen.net/gitweb/?p=rbraun/gnumach.git;a=shortlog;h=refs/heads/vm_policy_2g_dde
repository: git://darnassus.sceen.net/rbraun/gnumach.git
branch: vm_policy_2g_dde
orts regarding portability have been made there but I'm
almost positive having a semi or completely compatible interface would
help a lot). The Linux mechanism is also very well documented and
they've spent time on the tricky issues already, so we might just
avoid wasting time there.
In any case, it's definitely the right direction.
--
Richard Braun
h_port_deallocate is also available as a trap.
It's faster because it's futexes. The fact that there is a trap or not
only matters in the worst case, and personally I'd still do it with
mach_msg for consistency (trap based messaging is just a dirty
performance hack).
--
Richard Braun
an "operation" parameter. For gnumach, I think
> having separate RPC's is the better way.
Again, I agree, RPCs are operations. And again, if a simple wrapper can
be built on top, it's all fine.
--
Richard Braun
apparently. My guess is something got
wrong in the Linux glue code for in kernel drivers. Could you build
gnumach with --disable-net-group --disable-pcmcia-group
--disable-wireless-group and try again please ?
--
Richard Braun
> I don't think there is heavy contention actually :)
I agree. That's the point of futexes.
--
Richard Braun
On Sun, Feb 28, 2016 at 11:12:25PM +0100, Samuel Thibault wrote:
> Just sending my big +1 on this project :)
Same here.
I'd like to note that I'm personally pushing so that we have less
students but with better projects. This one definitely fits.
--
Richard Braun
> --disable-wireless-group and try again please ?
>
> Yes, that did it. The latest gnumach can be booted with GRUB when
> those options are disabled.
Thanks again for testing this. We've had several reports pointing in
that direction. I'll try to fix this soon and update the branch.
--
Richard Braun
On Mon, Mar 07, 2016 at 11:13:26PM +0800, 刘思远 wrote:
> The source code of my toy operating system is hosted on GitHub
> https://github.com/leasunhy/leanux
That's a nice little project you have there. Good luck for your
application.
--
Richard Braun
On Sun, Feb 28, 2016 at 03:50:06PM -0500, David Michael wrote:
> On Sun, Feb 28, 2016 at 3:37 PM, Richard Braun wrote:
> > On Sun, Feb 28, 2016 at 03:27:50PM -0500, David Michael wrote:
> >> The same GRUB has no problem booting older gnumach (bee3f0) or Linux.
> >> A
On Tue, Mar 08, 2016 at 12:00:03AM +0100, Richard Braun wrote:
> On Sun, Feb 28, 2016 at 03:50:06PM -0500, David Michael wrote:
> > Yes, that did it. The latest gnumach can be booted with GRUB when
> > those options are disabled.
>
> It seems we've been having this bug
when we have a good
user space replacement. As a result, I suggest we remove the rtl8139
driver from the kernel.
--
Richard Braun
ariables of our tests.
--
Richard Braun
--disable-rtl8139
and running KVM with either ne2k_pci, rtl8139 or pcnet, I'm unable to
reproduce any crash.
--
Richard Braun
#x27;s fine, but we'll discuss this on the GSoC web site once
student applications are open.
--
Richard Braun
you're going to present it.
And keep the list in copy.
--
Richard Braun
301 - 400 of 595 matches
Mail list logo
