Hello, is there any progress on this issue?
We have confirmed that it is an Alpine problem. We're not sure
exactly where the problem occurs (the usual suspect, grsec, has been
innocented), but we know it's related to the Alpine package build
and to virtualization.
In any case, the s6-rc code
Hello,
s6-2.6.0.0 is out.
It fixes a race condition that could lead
s6-svwait/s6-svlisten/s6-svlisten1
to miss back-to-back state changes. (Sorry about that.)
Unfortunately, the fix required change to some externally visible ABI,
so it enforces a major version bump.
External binaries
I finally managed to get execline to compile - turns out there was an
error in the options I had passed to configure:
I'm relieved. If it had been a bug, it would have been quite nasty
to solve. ;)
Thanks for the info.
--
Laurent
Just to be sure, I released skalibs-2.5.1.1, so there's a stable
version
with that bug fixed.
Update: the 2.5.1.0 bug breaks s6-rc-update pretty badly, so upgrading
to skalibs-2.5.1.1 is highly recommended.
--
Laurent
This change is intended. Laurent had never meant for s6-envdir to
ignore directories, the previous behaviour was just a coincidence.
When I made him asked him about this last week, he immediately made the
parsing code in skalibs stricter.
To be more precise: the fact that directories were
Hello,
Due to a datacenter closing in the future, the alyss.skarnet.org server
is migrating.
Expect some perturbations while I fix the inevitable problems that will
occur, and while the DNS caches are updating.
Sorry for the inconvenience - I'm doing my best to minimize it.
--
Laurent
Any chance you'd be willing to accept a patch like this to let the user
choose an -fstack-protector variant instead of hardcoding no? The
default behaviour is of course unchanged.
If you add a -fstack-protector option to CFLAGS, it will override the
default of no. Isn't that enough?
--
Yes, I obviously don't want to replace CFLAGS or CPPFLAGS whole-sale,
as
there might actually be necessary options in there, and I certainly
don't want to pick those out and have to provide them manually.
I'm currently making changes to the configure script and Makefile that
should have the
s6-rc: 0.0.2.1
s6: 2.2.4.3
skalibs: 2.3.9.0
Yeah, all bets are off with those versions - in particular the old
skalibs, a lot of bugs have been fixed since then.
I'll release a new s6-rc before next week; and if the s6-rc-init crash
happens again then I'll look into it.
--
Laurent
Hello,
skalibs, execline and s6 just received bugfix releases.
Nothing serious; the most important thing is probably that it's now
easier to append or override build-time flags via CPPFLAGS, CFLAGS and
LDFLAGS. Also, mkstemp() is used when appropriate, which allows more
functions to be used
Well, for skalibs in particular, that happens to work, since you (for
some reason) add -fno-stack-protector to CPPFLAGS, and CFLAGS happens
to come after in the compile rule. But for e.g. s6 and s6-rc, you
append fno-stack-protector to CFLAGS, so it doesn't matter what CFLAGS
contained at
I've at least twice have s6-rc-init crash due to an assert fail in
glibc. On the first occasion, I got
Thanks for the report. Will investigate as soon as I can.
--
Laurent
On a modern linux system, s6-rc-update ends up calling getrandom(2) via
the random_sauniquename() library function. But on embedded systems,
and
especially during early initialization, that is likely to block, which
means that the entire boot process hangs.
Yes. That's a feature. You
In the latest s6-rc git, s6-rc-update now creates its new livedir with
mkdtemp() and does not use any cryptographically strong random number
generation.
Please test it and tell me if it works for you.
--
Laurent
I think the update makes it clear enough to me, but I did not find any
update on the recompilation issue.
https://skarnet.org/software/s6-rc/upgrade.html ?
--
Laurent
This is certainly an inexcusable PEBKAC; however, the cause was not
really obvious from the error output of the s6-rc programs. So perhaps
we can make the error messages more indicative; or if that requires too
much additional complexiy, emphasise in the s6-rc documentation that
the
service db
I see; I think this version is OK, but the update seems to be absent
from the git repository?
Whoops! Sorry, fixed. :)
--
Laurent
Do you plan to do a bugfix release soon'ish?
I can do one some time later this week if you need it. However, if
it's not urgent, I'd like some time to investigate the s6-rc-init crash
you reported, if it's still happening for you. I cannot reproduce it
so far.
--
Laurent
Hello,
New releases of most skarnet.org packages are available; these are
mostly bugfix releases, but there are a few new features. skalibs
had to change an interface to fix a bug, which explains the major
version bump.
* skalibs-2.6.0.0
---
- Bugfixes, including a
src/librandom/random_makeseed.c says that that's precisely the point; I
was quoting verbatim.
Ah, indeed. My mistake. Full disclosure: at first I wanted to speed up
the /dev/urandom initialization. Then, after careful re-reading of the
Linux man pages, I realized it wouldn't work; but I found
Hello,
About two years ago, there was some talk on the Busybox mailing-list
about the need for a version of "mdev" that would not use a separate
process for every uevent (as a hotplug manager does) but that would act
as a daemon, able to handle a series of uevents - typically read from
the
I tried rewriting the whole thing in execline and while I'm pretty sure
it's doable it's not easy.
A direct translation of Casper's script to execline could be:
#!/command/execlineb -P
backtick LIST { cat /path/to/xyz.env }
importas -nsd"\n" LIST LIST
env -i ${LIST}
/path/to/xyz
It's not as
"Applications should not access sysfs" doesn't mean there should be
only one version of the intermediate library. There's already
Systemd-Udev and Eudev versions, at least. For the case of Xorg, it
just means sysfs access should not be hard-coded in the source of Xorg,
but there should be
Hello,
New releases of some skarnet.org packages are available.
Those are preparation features for future releases, bugfixes, or
compatibility improvements (for MacOS, for default-PIC toolchains).
* skalibs-2.6.2.0
* execline-2.3.0.4
* s6-2.6.2.0
* s6-rc-0.3.0.1
There's only one point
Hello,
The skarnet.org mailing-lists use a probing system to automatically
unsubscribe addresses that don't work anymore. When an address fails
3 consecutive times, it gets removed.
In the past year, or sometimes at an older time than that, a lot of
addresses have apparently failed enough
This kind of mass unsubscription is often due to DMARC.
Ah, thanks. Time for me to understand how DMARC works, then, and make
ezmlm work with it.
Sigh. E-mail gets more and more broken every year; soon it will just
become impossible to run a list.
--
Laurent
I noticed in the kernel log that the module was in fact correctly
loaded, but s6 was stuck waiting for something (I noticed this process
when doing a "ps"):
1213 root s6-ipcserverd -1 -- s6-ipcserver-access -v0 -E -l0 -i
data/rules -- s6-sudod -t 2000 -- /usr/libexec/s6-rc-oneshot-run -l
I fully understand that this list is focused on good programming
practice and libudev is considered something disgusting in that
respect. My question was triggered by the frustration, probably shared
by many, of not being able to replace Udev by Mdev or Mdevd, only
because Xorg autoconfig
Will there be a plan to add something roughly isomorphic to libudev?
Not in the short or medium term. The problem of libudev, and this
is shared by everything from freedesktop.org as well as systemd, is
that its design and structure cannot be achieved without a monolithic
implementation, and
I came across this http://skarnet.org/software/lh-bootstrap/. As I am
interested, I have looked for a x86_64-to-ramips toolchain
I don't have MIPS toolchains :/
If you know the exact gcc building flags needed to make one, I can
easily build one. And if you know the correct set of sysdeps to
I was thinking of other applications of NaCl, like encrypting files to
keep in a computer, rather than network stuff.
You can still use NaCl for this. But salted was all about networking.
--
Laurent
But I do have a question: as documented, I find that when stage1's
child
process execs into stage2, there is a devtmpfs mounted at /dev. I don't
understand where this happens! I see where the stage 1 script mounts a
tmpfs at /run, but I don't see anything there, or in the initial s6
scandir,
```
#!execline
define url example.com
s6-tcpclient $url 80
execlineb "
foreground {
fdmove 1 7 echo -en \"…\"
fdmove 0 6 cat
}
"
```
I have trouble understanding what you want to do in the general case,
because in this example, s6-tcpclient takes a whole command line, so
you do not have to
Thanks. It confirms what I had figured out (something's wrong in
the way gensetdyn_ready() computes its new size)
Well I just noticed you're linking against libskarnet.so.2.5.1.1.
As I said before, it is a *known bug* in skalibs-2.5.1.1, which has
been fixed in 2.6.0.0.
Please try with the
I was able to build the image for my target system and run it in qemu
and
still reproduce the problem. I first ran it through valgrind which
pinpoints a few possible issues:
http://steffe.net/s6/valgrind_nodebug.txt
Then I ran it through gdb and just did a "bt", let me know if I can
help
you
Thing is, before writing to that uevent, there was nothing of the sort
in /sys/dev.
What it looks like is the kernel not assigning a major and a minor
to the device before you manually trigger the "add". I suspect it's
just that the relevant module is not loaded, but why would that be
However, somewhat recently I believe I heard (read) you say on IRC that
the recommended way was actually the other way around, so I switched to
start netlinkd first, and then run coldplug.
Well, it's really up to you.
The thing is, if you coldplug first then listen to the netlink, you
have a
I was thinking about unit test to test your functions,
and building your code and installing it into a Docker container.
That's the very point: s6 is organized into several small
executables interacting together, so unit tests for functions are
pretty difficult to write. The units that would
Hello,
New releases of some skarnet.org packages are available, including
major version releases to s6 and s6-rc.
* skalibs-2.6.3.0
---
- Bugfixes.
- Lots of new functions, including the textclient library, which is
a simpler IPC library than skaclient for processes that
And to answer your question, about what could cause such a burst of
events, this time I have an easy answer: mdevd-coldplug !
Yes, I forgot to mention this.
The kernel-generated events seem to be significantly heavier than the
mdev-style synthetic ones. The DEVPATH is the full path without
Thanks! skalibs-2.6.3.1 released.
--
Laurent
> Would this difference explains why there is no redefinition linking
> error with musl, or am I missing something?
I don't know. I will ask people who know. It is strange indeed;
I didn't even notice that the strnlen sysdep was "no" with musl,
else I would have fixed it...
What is the
It might be worth to look at what systemd [1] and eudev [2] do here.
When 'udevadm trigger --type=devices --action=add' is performed after
the udev daemon starts executing, it looks like what happens is
basically this:
Thanks a lot for studying this. That's what I was going to do, but
I
I've found the command intuitive but the documentation confused me a
little bit.
So here's a patch.
ISTR you actually submitted the same thing a few months ago and I did
not react. Sorry, I should have replied sooner.
The documentation is correct and I won't apply the patch. But it's
not
`execline` does indeed exits when not given arguments, and displays the
error `execlineb: warning: too few arguments: expecting at least 1 but
got 0` but the first command gets run.
```
$ execlineb -S1 -c 'echo wrong'
execlineb: warning: too few arguments: expecting at least 1 but got 0
wrong
#!/bin/execlineb -W -S 1
You're running into an unfortunate side effect of shebang lines where
only the first argument is available.
Exactly. Shebang parsing is done by the kernel, which does not have
a complete command line parser. (If it had, the execlineb program
would be redundant: we
Hello,
New releases of some skarnet.org packages are available, as well
as a new package.
* skalibs-2.7.0.0
---
This is a major version release of skalibs, so compatibility with
previous versions is not guaranteed. Make sure to also upgrade the rest
of the skarnet.org
Oh. That's kind of drastic. I re-read the documentation of nsss and
utmps, in case I missed it the first time, and still think none of it
suggests that this is the intended behaviour.
You're right, the documentation should mention that libc header files
will be replaced. I'll add something.
That brings me to another question: is there a collection of
more execline utilities somewhere? I suspect the standard
GNU coreutils are not quite to your liking in a lot of cases.
The execline package has all the execline-specific utilities you
need to perform execline scripting (although if
So I’d like to have an ifthenelse and return the
same envvar from both branches.
https://www.mail-archive.com/skaware@list.skarnet.org/msg00311.html
--
Laurent
I haven’t found a way to print a file to stdout with either
execline nor s6-portable-utils. The `s6-cat` utility only
echoes stuff that is already coming from stdin,
and unlike the shell, execline doesn’t have `<`.
https://skarnet.org/software/execline/redirfd.html
"redirfd -r 0 $file
elglob -0 fs somedir/*
if { test -n $fs }
ln -t otherdir $fs
the test will fail if there’s more than one file in `somedir`.
Is there a way to put a split variable into one variable again?
Your next door "echo" command will do just that (or s6-echo
if you risk having dashes and want reliable
I intend to package skarnet softwares in Debian. When packaging
execline, I find it produces a lot of binaries with common names. Thus
I want to put them in path like /usr/lib/execline/.
Do not do this.
Several programs, including s6 and s6-rc, rely on execline binaries
to be in their PATH.
Thanks for this note, I haven't looked at these packages. At first I
thought execline binaries will only be called inside execlinep
scripts. Could you give some examples that s6 calls these binaries
directly except exelinep?
In addition to what Casper said: for instance, some of the
> [1] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=906250#22
That's an incredibly disingenuous argument. Debian isn't POSIX-
compliant by any stretch of the imagination, and they currently
don't provide POSIX-compliant cd/umask/wait binaries because they
have correctly identified that
However the binaries(cd, umask, wait) that cause problem are not
dependency of s6.
Those binaries are not causing problems.
What *is* causing problems is fellow developers who *think* those
binaries are problematic, for reasons that all come from ignorance
of how Unix (and the shell) works.
So, I noticed the package includes example OpenRC service scripts
(wow!) for utmps-utmpd and utmps-wtmpd. Probably not the kind of
testing you hoped for, but for the sake of correctness: they won't
work :-P
On the contrary, that's exactly the kind of testing I appreciate. :)
I've managed to do
Applied, thanks!
--
Laurent
Ok, so let me ask you this: I want to delete my avltree. That is, I'd
like to remove all its nodes and be done with it, i.e. I need to free
all memory associated with it. I guess avltree_free() is enough, no
need to delete each node individually?
Yes, avltree_free() is enough, because an
- When using avltree_iter() we need to pass a avliterfunc_t_ref, which
takes 3 arguments. If I understand correctly, first is the uint32_t
index, then the tree's "internal" (unsigned int) index, and my
pointer.
"The uint32_t index" is confusing terminology, so to clarify, I call
this number
Hello,
execline-2.4.0.0 is out.
As you all know by now, the main obstacle to adoption of skaware by
distributions and mainstream Unix folks is the user interface; so, the
obvious next improvements to skarnet.org software is to provide
interfaces people are comfortable with. This is true for
Hello,
New releases of some skarnet.org packages are available.
* skalibs-2.6.4.0
---
- Bugfixes.
- Several new functions, including a small FIFO library for generic
object queues (genqdyn.h).
https://skarnet.org/software/skalibs/
git://git.skarnet.org/skalibs
* A
Hello,
execline-2.5.0.0 is out.
It is with a very heavy heart that I must do this release.
User reports have come in by the hundreds and they are almost
unanimous (sorry, Colin): they don't like the 2.4.0.0 change,
pretending it hurts readability (as if), and writability too,
of execline
My reaction:
:) :)
* "Wait, did he rename the C source files too? Like
src/execline/=.c, src/execline/;.c, etc.?"
I tried. Really, I tried. The filesystem didn't have any problem
with that, nor did shell scripts, provided the appropriate amount of
quoting was given; but *make* choked.
* s6-2.7.1.0
--
Make that s6-2.7.1.1, since a stupid (non-critical, but annoying if
you prefer your warnings to be displayed when something goes wrong
rather than when things work normally) made it through. ^^'
--
Laurent
But I probably will not install binaries like `if, exec, wait` to
default PATH(like /usr/bin). Although I understand this is possible
technically(and currently no other programs claim these names).
The actual reason is that I can't afford time to argue with other
developers. When I ask on
The existence of my toolchains, and Zach's, is relatively independent
from the existence of lh-bootstrap and mkroot.
As far as I'm concerned, the creation of these toolchains was an
exercise in working with musl-cross-make at start - and then some
people found them useful, so I built them
Probably a dumb question, but is there a reason there is no
`cp` and no `mv` in `s6-portable-utils`?
Because reimplementing a full POSIX utility suite isn't a goal.
There are already a lot of those you can choose from (coreutils,
busybox, toybox, sbase+ubase, etc.)
That said, if you don't
The next question was the long-term strategy with systemd: we have 2
approaches
1) mix systemd (PID 1) to use the boot phase (especially one shots)
and later (after system initialization) use s6 to manage all services.
2) just use s6
My question to the list is, what do you use today? Do you
Our view is, that using an non-mainstream distribution is a lot of work
for our purpose of being an PaaS provider. That would in the end mean
packaging a lot of services on our own.
Today the mainstream distribution doesn't package anymore all needed
services in the needed versions. In practice
This fixes a segfault when a user calls uint320_fmt(NULL,...),
and makes the fmtlist functions return the correct number of
characters when called with NULL.
Applied, thanks!
--
Laurent
Hi all - I just wanted to file a quick report.
I tried cloning down the skalibs git repo to my macOS box, which uses
the default, case-insensitive filesystem.
Since there's both mkltemp.c and mkLtemp.c, I immediately ran into
problems with git reporting modifications I hand't made, etc.
Ah,
Adelie (musl-posix) must be getting really close to replacing
sysv/openrc with full s6.
sysv is done: on Adélie, you can now choose between sysvinit and
s6-linux-init as your init system. Both are supported.
Providing a packaged alternative to OpenRC, however, will be much
more difficult
Short version: For recent libc releases, and at least on Gentoo,
getgrent() and endgrent() seem to magically set errno to EINVAL (I
think), except when errno's value is actually meaningful.
> (...)
End of file or error (errno = Success)
errno = Invalid argument
POSIX says:
But it turns out that, with this configuration, __nss_endent() *also*
wants to call the implementation of endgrent(3) from each of these
services. And the one from libnss_db.so, named _nss_db_endgrent(), is
just a wrapper around a munmap(2) system call, via an intermediate
internal_endent()
Hello,
Following some UX reports in the #s6 IRC channel, a new version of
the s6-linux-init package is out.
* s6-linux-init-1.0.1.0
-
- Binaries have less restrictive permissions. Root-only binaries
are now readable (though still not executable) by normal users.
-
Hello,
Some minor updates are available.
* skalibs-2.8.1.0
---
- Small bugfixes.
- A new wrapper, skalibs_regcomp(), has been provided as a portable
replacement for regcomp(). Some systems (looking at you again, BSDs)
do not accept empty regular expressions and regcomp()
Applied, thanks.
I didn't bother before because a make build is inherently
non-deterministic, unless the author manually defines a total order
on all the prerequisites of a target. I suspect the implementation of
GNU make to implicitly define such an order, but there's nothing of
the sort in
1. After resuming from suspension or hibernation, s6-log no longer shows
the correct timestamp. Can this be fixed? Thanks.
This is a much more complicated issue than it sounds.
Short answer: no, the system needs to readjust the system clock when
resuming from suspension/hibernation; it's
Thanks Guillermo. I should have thought about this. So Linux is doing
the right thing (fixing CLOCK_REALTIME on resume), and the problem
only appears with CLOCK_MONOTONIC. It is indeed one of the drawbacks
of using --enable-monotonic when you need a wall clock. And it is a
skalibs problem that
Doesn't s6-log (indirectly) use clock_gettime() with a CLOCK_MONOTONIC
argument if skalibs was built with --enable-monotonic?
Yes, it does. But --enable-monotonic is a bad idea for long-lived
processes that need timestamping. I need better clock interfaces
in skalibs, the current ones don't
On 9/14/2019 2:11 PM, Jens Rehsack wrote:
I don't have such a database and such a database is imposible to have.
That's a real shame, because it's the *only way* cross-compilation
can ever work.
You praise autotools for supporting ac_cv_$check variables, that can
be supported from the
On 9/17/2019 7:37 PM, eadwar...@disroot.org wrote:
Posix says about close(2):
"If close() is interrupted by a signal that is to be caught, it shall return
-1 with errno set to [EINTR] and the state of fildes is unspecified."
Tell me about it...
The fact that close() can be
s6-linux-init-maker creates an .s6-svscan/SIGINT script that reboots
the computer, and Adélie Linux' packaging of s6-linux-init-1.0.2.0
appears to contain an /etc/sysctl.d/ctrlaltdel.conf file that says
"kernel.ctrl-alt-del = 0". So one might think that having the kernel
send a SIGINT signal to
if (WIFSIGNALED(wstat))
//...
else if (WEXITSTATUS(wstat))
//…
else if (WEXITSTATUS(wstat))
//…
Good catch, thanks!
Fixed in the latest git. Bugfix release coming at some point in the
future. (I'm not hurrying because it's only a display bug.)
--
Laurent
What I *can* do, though, is change the skalibs build system so that
on cross-compilation, you only need to provide the few sysdeps that
cannot be probed at all.
I spent today on this, and managed to bring down the necessary amount
of manually provided sysdeps to one (1).
The skalibs git head
Has anyone else gotten such a warning from the listserv?
Can't do much about it, sorry. ezmlm comes from a time when e-mail
was simpler and less broken (or, at least, broken in different ways
from today).
Your server is rejecting some mails for reasons unknown. Likely a
misconfiguration. Or it
Since I'm not using static libraries, I would not expect to need the
"--with-lib" configure option.
The --disable-static flag only tells the build system not to
build the static execline library. By default, executables are still
built against the static version of skalibs.
If you want to
/opt/local/bin/s6-svscan: error while loading shared libraries: libs6.so.2.9:
cannot open shared object file: No such file or directory
Is this because the '-rpath' linker option is not used at compile time
to add the library directory to the runtime library search path?
You could say that -
Makefile uses CC instead of $(CROSS_COMPILE)$(CC),
So CC shouldn't strip cross prefix when configure.
That sounds right, but ISTR testing it when I made the change and
it worked when cross-compiling. There must have been something else
going on; I will test tomorrow and possibly apply.
So, whoever calls s6_supervise_lock_mode() should correctly die, as
there is an actual error (that should be reported to the Hurd
developers); libs6, libskarnet and the libc are just the messengers.
Commenting out the errno check just sweeps the error under the carpet.
I assumed the OS was
Confirmed; I must have been confused when testing earlier.
Fix pushed, thanks Shengjing.
Bugfix releases coming soon.
--
Laurent
Since a new release is coming soon, is there any chance to apply
another cross compile patch[1]?
Applied, thanks for the reminder. New versions pushed, release
announcement coming soon.
--
Laurent
Hello,
New versions of all the skarnet.org packages are available.
This is mostly a bugfix release, as shown by the *.1 version numbers.
It just squashes a couple build bugs that prevented proper
cross-compilation.
If you are not cross-compiling skarnet.org packages, you can mostly
ignore
Excellent analysis, as always. :) Thank you.
* If GNU libc is used, a 'choose cl' test is unreliable.
* GNU/Hurd and GNU/k*BSD would be 'getrandom: no' OSes, at least until
the libc is updated to make a system call on GNU/kFreeBSD [1]
* A 'choose clr' test would be reliable.
* Any other
Overridable for the win ;)
Are we giving advice again without checking first, hmmm?
./configure --with-sysdep-getrandom=no <-- Try it. :P
My point is that ld *should* be a reliable way to autodetect those
things, and I want to make sure it works too.
--
Laurent
While building skalibs-2.9.1.0, I notice that the getrandom option isn't
available on FreeBSD 12.1 Stable. What I have from ./configure --help is:
"...
Sysdeps autodetection override:
--with-sysdep-K=V assume sysdep K has the value V
[autodetected]
List of mandatory K for
If compiler predefined macros like __linux__ or __NetBSD__ are OK for
, how about #if defined(__linux__) ||
defined(__GNU__)? GCC predefines both __GNU__ and __gnu_hurd__ (I
think, can be checked with 'cpp -dM -
Do you mean __GNUC__ instead of __GNU__ ?
Testing defined(__GNUC__) would yield
__gnu_hurd__
That's perfect. I replaced __GLIBC__ with this. So skalibs/nonposix.h
will define _GNU_SOURCE on Linux and Hurd, i.e. the systems whose
function visibility benefits the most from it.
Testing kernel ftms is generally ugly, but it's the right thing for
nonposix.h whose goal is
The errno I got on Hurd is `EIEIO 1073741928 Computer bought the
farm`... Seems a Hurd specific errno...
That is really non-conformant, and deserves to be reported to the Hurd
team.
What is happening is that "control" is a named pipe, and the attempted
operation is open(control, O_WRONLY |
201 - 300 of 534 matches
Mail list logo
