Hi,
I noticed this today and commited your change.
mbuhl
On Tue, Jun 02, 2020 at 12:33:36PM +, Yifei Zhan wrote:
> Hi,
>
> Several links on loongson.html are now broken as zkml.lemote.com is no
> longer being resolved.
>
> I think it's a good idea to redirect them to their archived
Dear tech,
while reading through ffs code I was looking for calls to ffs_truncate.
This would have saved me a few seconds.
OK?
mbuhl
Index: ufs/ffs/ffs_vfsops.c
===
RCS file: /cvs/src/sys/ufs/ffs/ffs_vfsops.c,v
retrieving revision
On Mon, Apr 03, 2023 at 05:53:49PM +, Klemens Nanni wrote:
> 03.04.2023 15:03, Klemens Nanni пишет:
> > This means no behaviour change for plain installs, but working boot for
> > softraid installs.
> >
> > 'cvs diff -b -U0':
> >
> > @@ -39 +39 @@ md_installboot() {
> > - local
Below is an updated diff addressing the following concerns:
On Fri, Mar 17, 2023 at 03:45:59PM +0100, Alexander Bluhm wrote:
> The FreeBSD Code puts the check_nifree label before the if
> (fs->fs_cs(fs, cg).cs_nifree == 0) check. Could you move this chunk
> a little up to keep us in sync?
On
Dear tech,
I have a machine with some kind of hardware defect.
Smartctl shows that one SSD regularly has an unexpected power loss
(Attribute 174):
174 Unknown_Attribute 0x0032 100 100 000Old_age
Always - 284.
On the console I see the following output:
root on sd0a
Hi gerhard and tech,
You wrote the following in 2015 and I was wondering if it was worth
upstreaming it:
The PPPoE kernel module gets a new option auto for the authproto
in addition to pap and chap in client mode.
In this case the authentication protocol sent by the server in the
LCP configure
Dear tech@,
codechecker found the following problem with fsb in sbin/mountd:
mntsrv(...)
...
struct statfs fsb;
...
if (realpath(rpcpath, dirpath) == NULL) {
bad = errno;
if (debug)
Dear tech,
Looking into various reports on errors with mini routers that use
igc(4), I found the following diff in FreeBSD:
https://github.com/freebsd/freebsd-src/commit/29d7f1ff579579711dd5a3325480728b8ed45f8c
I additionally deleted the igc_set_d0_lplu_state_i225 and
igc_set_d3_lplu_state_i225
Dear tech,
currently the MAP_CONCEAL flag for mmap(2) can be bypassed with a
stack overflow. FreeBSD introduced a similar flag called MAP_NOCORE.
They also save the auxinfo pointer in some struct that is attached
to the proc struct, similar to the diff below.
Here is an example CTF challenge:
On Thu, Nov 10, 2022 at 01:29:13AM +0100, Theo Buehler wrote:
> On Thu, Nov 10, 2022 at 01:10:51AM +0100, Moritz Buhl wrote:
> > errstr is never set but fail does:
> > RSA_meth_free(rsae_method);
> > fatalx("%s: %s", __func__, errstr);
> &
errstr is never set but fail does:
RSA_meth_free(rsae_method);
fatalx("%s: %s", __func__, errstr);
Found by codechecker.
OK?
mbuhl
Index: ca.c
===
RCS file: /cvs/src/usr.sbin/relayd/ca.c,v
retrieving revision 1.39
The same code is in httpd but there it was fixed in
commit 6b535b529336a3fd1beb56c42ff5755b84ba9b03
Author: jung
Date: Sun May 22 19:19:21 2016 +
fix unbalanced va_start and va_end macros
from Hiltjo Posthuma
"do." deraadt
Found by codechecker
OK?
mbuhl
Index:
Hi tech,
g is not freed in this error case.
Found by codechecker.
OK?
mbuhl
Index: usr.sbin/acme-client/netproc.c
===
RCS file: /cvs/src/usr.sbin/acme-client/netproc.c,v
retrieving revision 1.31
diff -u -p -r1.31 netproc.c
---
Hi tech,
In case send or recv fail in send_command, reply.error is stack
garbage. This wouldn't be possible if reply was zeroed. But
checking for ret == 0 ensures that reply was fully written.
OK?
mbuhl
Index: usr.sbin/apm/apm.c
In case lh_OBJ_NAME_insert returns NULL due to a failed malloc, onp
is leaked in OBJ_NAME_add.
Found by CodeChecker.
OK?
mbuhl
Index: lib/libcrypto/objects/o_names.c
===
RCS file: /cvs/src/lib/libcrypto/objects/o_names.c,v
Hi Greg, Hi Joe,
dlg@ hinted to me that the ring might overwrite it's own starting
position with the current code.
Does this help?
mbuhl
Index: dev/pci/if_igc.c
===
RCS file: /cvs/src/sys/dev/pci/if_igc.c,v
retrieving revision 1.9
Hi tech,
Dear claudio,
ctx might leak due to a prefix/rib dump new/subtree failing in calloc
and then going to nomem in rde_dump_ctx_new.
I am wondering if a similar fix is missing in rde_dump_done after
the nomem label.
thoughts?
mbuhl
Found by CodeChecker.
Index: usr.sbin/bgpd/rde.c
Hi tech,
Dear claudio,
there could be an uninitialized stack memory access in pfkey_reply.
It looks like this:
struct sadb_msg hdr, *msg;
...
do {
rv = pfkey_read(sd, );
if (rv == -1)
return (-1);
} while (rv);
Looking for OKs.
On Sat, Oct 15, 2022 at 10:01:53PM +0200, Moritz Buhl wrote:
> With the previous diffs I am seeing sporadic connection problems in tcpbench
> via IPv6 on Intel 82546GB.
> The diff was too big anyways. Here is a smaller diff that introduces
> checksu
With the previous diffs I am seeing sporadic connection problems in tcpbench
via IPv6 on Intel 82546GB.
The diff was too big anyways. Here is a smaller diff that introduces
checksum offloading for the controllers that use the advanced descriptors.
I tested this diff on i350 and 82571EB, receive,
Another mistake
On Wed, Oct 12, 2022 at 01:45:30PM +0200, Moritz Buhl wrote:
> On Tue, Oct 11, 2022 at 04:16:15PM +0100, Stuart Henderson wrote:
> > On 2022/10/11 15:03, Moritz Buhl wrote:
> > > Here is a new diff for checksum offloading (ipv4, udp, tcp) for em(4).
> > &g
On Tue, Oct 11, 2022 at 04:16:15PM +0100, Stuart Henderson wrote:
> On 2022/10/11 15:03, Moritz Buhl wrote:
> > Here is a new diff for checksum offloading (ipv4, udp, tcp) for em(4).
> >
> > The previous diff didn't implement hardware vlan tagging for >em82578
> > w
Here is a new diff for checksum offloading (ipv4, udp, tcp) for em(4).
The previous diff didn't implement hardware vlan tagging for >em82578
which should result in variable ethernet header lengths and thus
wrong checksums inserted at wrong places.
The diff below addresses this.
I would
On Fri, Sep 16, 2022 at 11:38:49PM +0200, Moritz Buhl wrote:
> Syzkaller found a panic with sysv semaphores:
> https://syzkaller.appspot.com/bug?id=f7e8e2822779918d7a23d9ff9d7c0a3779c00a46
The last diff would introduced two memory leaks, seminfo.semopm is
the only seminfo value that can
Syzkaller found a panic with sysv semaphores:
https://syzkaller.appspot.com/bug?id=f7e8e2822779918d7a23d9ff9d7c0a3779c00a46
The problem is that the code uses a few globals (sema, semtot,
seminfo) that can change during sleeping points where the kernel
lock is released.
Usually copyin, copyout,
On Wed, Aug 31, 2022 at 05:44:31PM -0900, Philip Guenther wrote:
> kdump.c will need at least a SYS_recvmmsg line in the big table, and if you
> do a ktrmmsghdr() bit in the kernel a matching decoder will be needed in
> kdump.
Here is a new diff for kdump.
OK?
mbuhl
Index: usr.bin/kdump/kdump.c
On Tue, Sep 06, 2022 at 04:00:39PM +0200, Moritz Buhl wrote:
> Hi,
> here is the most recent diff for the libc part of send and recvmmsg.
> This requires a libc minor bump and therefore should be coordinated
> after snapshots are building normally again.
>
> To my understand
.Ox 7.2 .
Index: lib/libc/sys/w_recvmmsg.c
=======
RCS file: lib/libc/sys/w_recvmmsg.c
diff -N lib/libc/sys/w_recvmmsg.c
--- /dev/null 1 Jan 1970 00:00:00 -
+++ lib/libc/sys/w_recvmmsg.c 6 Sep 2022 09:42:08 -
@@ -0,0 +1,32 @@
+/
Here is an updated version of the kernel part for sendmmsg.
mbuhl
Index: sys/kern/syscalls.master
===
RCS file: /mount/openbsd/cvs/src/sys/kern/syscalls.master,v
retrieving revision 1.230
diff -u -p -r1.230 syscalls.master
---
I addressed your concerns as well as these of jca, just the kernel
part (and the new ktrace stuff) below.
One minor thing: I didn't see any kdump output where one struct was
contained in another one but I am printing it like ddb would so I
guess it should be fine.
Index: kern/syscalls.master
I created a pull request on their github:
https://github.com/NLnetLabs/nsd/pull/231
On Tue, Aug 30, 2022 at 04:33:10PM -0600, Todd C. Miller wrote:
> On Wed, 31 Aug 2022 00:19:20 +0200, Moritz Buhl wrote:
>
> > On Tue, Aug 30, 2022 at 10:59:43PM +0200, Claudio Jeker wrote:
> >
On Tue, Aug 30, 2022 at 10:59:43PM +0200, Claudio Jeker wrote:
> And nsd in base. It seems unbound does not use recvmmsg.
After 'make -f Makefile.bsd-wrapper config' recvmmsg is picked up.
The config compile test currently defines NONBLOCKING_IS_BROKEN
because of a missing include. Then vlen
+.Ox 7.2 .
.Sh CAVEATS
Calling
.Fn recvmsg
Index: lib/libc/sys/w_recvmmsg.c
===
RCS file: lib/libc/sys/w_recvmmsg.c
diff -N lib/libc/sys/w_recvmmsg.c
--- /dev/null 1 Jan 1970 00:00:00 -
+++ lib/libc/sys/w_recvmmsg.c 22 Apr 202
ok mbuhl
On Tue, Aug 30, 2022 at 05:13:35PM +0200, Sebastian Benoit wrote:
> ptype is never used.
> ok?
>
> diff --git sys/dev/pci/if_ix.c sys/dev/pci/if_ix.c
> index cb233034d23..72a221b97d9 100644
> --- sys/dev/pci/if_ix.c
> +++ sys/dev/pci/if_ix.c
> @@ -148,7 +148,7 @@ void
On Sun, Aug 07, 2022 at 12:41:29AM +0200, Alexander Bluhm wrote:
> Hi,
>
> If interface drivers have enabled transmit offloading for the payload
> checksum , IPv6 fragments contain invalid checksum. For fragments
> the protocol checksum has to be calculated before fragmentation.
> Hardware
Dear tech@,
the scan-build(1) tool from the llvm port on sbin/pfctl reported a
memory leak in the error path of load_feedback_profile.
There are more places that could be investigated.
Would it make sense to look into these?
Independent of that I was wondering if it makes sense to warn if
On Thu, Jul 21, 2022 at 10:25:09PM +0100, Jason McIntyre wrote:
...
> it looks like the "set limit" text in pf.conf(5) might need some small
> adjustments:
>
> - as well as the "anchors" limit, it does not document "pktdelay-pkts"
>
> - for entries where defaults are not documented, it is not
On Mon, Jun 27, 2022 at 08:07:32AM +1000, Jonathan Gray wrote:
> On Sun, Jun 26, 2022 at 04:43:59PM +0200, Moritz Buhl wrote:
> > On Sun, Jun 26, 2022 at 12:23:58PM +0200, Moritz Buhl wrote:
> > > Hi,
> > >
> > > I noticed that for some offloading-capable em c
Hi tech,
for the other two DIOCX ioctls syzkaller showed that it is possible
to grab netlock while doing copyin.
The same problem should exist for DIOCXCOMMIT but syzkaller didn't
find it yet.
In case anybody can reproduce the witness lock order reversals the
syzkaller can produce, the diff
On Tue, Jul 19, 2022 at 10:58:08PM +0200, Alexander Bluhm wrote:
> On Tue, Jul 19, 2022 at 07:18:54PM +0200, Moritz Buhl wrote:
> > A solution would be to move the allocation of the pf_anchor struct
> > into a pool. One final question would be regarding the size of the
> &g
Dear tech@,
I am investigating a syzkaller reproducer found in the
"no output from test machine (7)" crashes:
https://syzkaller.appspot.com/bug?id=d93e92fde3857c69df2cf46b4244d9814c4318a7
https://syzkaller.appspot.com/text?tag=ReproC=116ee2e008
The code calls DIOCXBEGIN, with different
Hi tech@, Dear miod@, Dear kettenis@,
On arm the lib/libm/rint regress test is failing. After adding some
printf debugging into libm it turns out that the problem is the
cast from double to long long.
http://bluhm.genua.de/regress/results/2022-07-12T04%3A17%3A03Z/logs/lib/libm/rint/make.log:
On Sun, Jun 26, 2022 at 12:23:58PM +0200, Moritz Buhl wrote:
> Hi,
>
> I noticed that for some offloading-capable em controllers checksum
> offloading is still disabled and I couldn't find a reason for that.
>
> It was assumed during initial implementation:
> https://marc
Hi,
I noticed that for some offloading-capable em controllers checksum
offloading is still disabled and I couldn't find a reason for that.
It was assumed during initial implementation:
https://marc.info/?l=openbsd-tech=137875739832438=2
According to the datasheets of the i350 controllers listed
On Tue, May 03, 2022 at 09:31:51PM +0200, Alexander Bluhm wrote:
...
> The code is too complex to be sure what the reason of the syzkaller
> panic is. Sleep in malloc is correct anyway and may improve the
> situation.
>
> Functions with argument values 0 or 1 are hard to read. It would
> be
Dear tech@,
the following diff should implement vlan hwtagging for igc.
I would appreciate feedback from further testing.
OK?
mbuhl
Index: sys/dev/pci/if_igc.c
===
RCS file: /cvs/src/sys/dev/pci/if_igc.c,v
retrieving revision 1.9
On Wed, May 25, 2022 at 04:18:00PM +0200, Moritz Buhl wrote:
> Dear tech@,
>
> The attached diff should handle checksum offloading for igc(4)
> for IPv4, TCP, and UDP.
>
> I tested with "Intel I225-LM" rev 0x03,
> kevlo@ tested this with "Intel I225-V"
Dear tech@,
The attached diff should handle checksum offloading for igc(4)
for IPv4, TCP, and UDP.
I tested with "Intel I225-LM" rev 0x03,
kevlo@ tested this with "Intel I225-V" rev 0x03
OK?
Index: /sys/dev/pci/if_igc.c
===
RCS
Hi tech@,
Syzkaller found a few crashes in pf_anchor_global_RB_REMOVE.
https://syzkaller.appspot.com/bug?id=a97f712331903ce38b8c084a489818b9bb5c6fcb
and also
https://syzkaller.appspot.com/text?tag=CrashLog=15ace9aaf0
The call stack is something like this:
pf_anchor_global_RB_REMOVE
Hi tech@,
syzkaller reported pfi_get_interfaces
https://syzkaller.appspot.com/bug?id=963ef41930bb84af584be3bf910a3f67d65581a0
which does a copyout while holding NET and PF lock.
The following diff should address this.
mbuhl
Index: sys/net/pf_if.c
On Tue, Apr 19, 2022 at 09:00:08PM +0200, Anton Borowka wrote:
> Hi,
>
> the attached diff fixes 2 issues with line wrappings in wall(1).
...
I can confirm that your patch fixes the described problems. But I
am wondering if it makes sense to split the ch == '\n' case and the
cnt == 79 case into
send
function call appeared in
.Bx 4.1c .
+The
+.Fn sendmmsg
+syscall first appeared in Linux 3.0, was reimplemented for
+.Nx 7.0 ,
+and ported to
+.Ox 7.1 .
Index: lib/libc/sys/w_recvmmsg.c
===
RCS file: lib/libc/sys/w_recvmmsg.c
di
Hi,
Since 2015 pf_rollback_rules always returns 0. Make it void.
OK?
mbuhl
Index: sys/net/pf_ioctl.c
===
RCS file: /cvs/src/sys/net/pf_ioctl.c,v
retrieving revision 1.376
diff -u -p -r1.376 pf_ioctl.c
--- sys/net/pf_ioctl.c 4 Apr
Any insights?
On Mon, Jan 10, 2022 at 03:18:47PM +0100, Moritz Buhl wrote:
> Hi tech@,
>
> The return value of add_child_sensors is returned in add_sdr_sensor,
> which is called by get_sdr. get_sdr mallocs psdr and only frees it
> if add_sdr_sensor returns 0. The assumpti
I thought I'd walk through my understanding of the code
to make it easier to review.
from sys/dev/ipmi.c:
1019 int
1020 get_sdr(struct ipmi_softc *sc, u_int16_t recid, u_int16_t *nxtrec)
1021 {
...
1046 psdr = malloc(sdrlen, M_DEVBUF, M_NOWAIT);
1047 if (psdr == NULL)
1048
Hi tech@,
https://github.com/openbsd/src/commit/0ea6bae06233cd25645df14602c3eda6bdff7dca.patch
the patch forgot to add mrep to the info struct, nfsm_dissect could
pass info.nmi_mrep to m_freem, which is currently uninitialized
stack memory.
Index: sys/nfs/nfs_subs.c
Hi tech@,
The return value of add_child_sensors is returned in add_sdr_sensor,
which is called by get_sdr. get_sdr mallocs psdr and only frees it
if add_sdr_sensor returns 0. The assumption is that psdr is attached
to a list in add_child_sensors otherwise. This is not the case if
the malloc
Dear tech@,
given the following Makefile:
# REGRESS_SETUP vs REGRESS_SETUP_ONCE example
setup:
@echo setup
setup_once:
@echo setup_once
test1:
@echo test1
test2:
@echo test2
cleanup:
@echo cleanup
REGRESS_SETUP = setup
REGRESS_SETUP_ONCE =
Any comments?
July 22, 2019 8:23 PM, "Moritz Buhl" wrote:
> Hi,
>
> here is another POSIX compatibility change I found while running
> NetBSD system call tests:
>
> The incompatibilities can be reproduced with the followding comands:
> `mv / foo`
> mv: rename
Hi,
some ssh unittest ist failing because it links stuff together in another way.
basically add 'sshbuf-misc.c' to SRCS.
additional suggestion: make variable assignments similar to ports
mbuhl
Index: regress/usr.bin/ssh/unittests/misc/Makefile
Hi,
some experts forgot to run the rpki-client regression tests after some
changes. The tests are relinking parts of the source and this is not great
but after all I still prefer some running tests. Patch attached.
thanks,
mbuhl
Index: regress/usr.sbin/rpki-client/test-cert.c
Hi,
the same as to unveil(2) also counts for realpath(3).
https://marc.info/?l=openbsd-bugs=156469573812165=2
Similar diff attached.
tests are here:
https://github.com/moritzbuhl/realpath-unmount-regress
thanks,
mbuhl
Index: sys/kern/vfs_syscalls.c
Hi,
here is another POSIX compatibility change I found while running
NetBSD system call tests:
The incompatibilities can be reproduced with the followding comands:
`mv / foo`
mv: rename / to foo: Is a directory
`man 2 rename` mentions EISDIR:
[EISDIR] to is a directory, but from is not
> Moritz, can you create a man page ERRORS diff?
Is the attached diff ok?
mbuhl
Index: lib/libc/sys/msgsnd.2
===
RCS file: /cvs/src/lib/libc/sys/msgsnd.2,v
retrieving revision 1.19
diff -u -p -r1.19 msgsnd.2
---
Hi,
while running some ported tests I noticed that msgsnd(2) did not error
when passing a message with mtype < 1, even tho the manual page states:
> mtype is an integer greater than 0 that can be used
POSIX says:
https://pubs.opengroup.org/onlinepubs/9699919799/
[EINVAL]
The value of msqid
Hi,
I made the FreeBSD msun regression tests compile on OpenBSD.
https://github.com/moritzbuhl/msun-regress
3 out of 19 test files pass. 14 files die after the first error case.
Two files (ctrig_test.c and trig_test.c) use atf and after some hacks
they report all error cases. 840 for ctrig_test
Hi,
while running some NetBSD syscall tests I noticed that listen(2) returns
the wrong errno when a socket is already connected. man 2 listen already
mentions EINVAL.
NetBSD bug:
http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=46150
NetBSD patch:
Hi,
while porting some NetBSD syscall tests to OpenBSD I noticed that the
getgroups test is failing. Simply put:
gid_t gidset[NGROUPS_MAX];
getgroups(-1, gidset);
This was fixed on NetBSD 8 years ago:
http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/kern/kern_prot.c
> if (SCARG(uap,
Hi,
as promised a patch to relayd to enable formatting warnings for yyerror.
Greetings,
Moritz Buhl
Index: usr.sbin/relayd/parse.y
===
RCS file: /cvs/src/usr.sbin/relayd/parse.y,v
retrieving revision 1.238
diff -u -p -r1.238
. A patch will follow.
Thanks for feedback and comments,
Moritz Buhl
Index: sbin/pfctl/parse.y
===
RCS file: /cvs/src/sbin/pfctl/parse.y,v
retrieving revision 1.696
diff -u -p -r1.696 parse.y
--- sbin/pfctl/parse.y 8 May 2019 21:31:30
egress/lib/libc/sys/atf-c.h27 Jun 2019 19:07:52 -
@@ -0,0 +1,66 @@
+/* $OpenBSD$ */
+/*
+ * Copyright (c) 2019 Moritz Buhl
+ *
+ * Permission to use, copy, modify, and distribute this software for any
+ * purpose with or without fee is hereby granted, provided that
Hi,
while testing arm hardware on OpenBSD I noticed that some floating point
operations cause failures of other tests.
In fact the current libm is incorrect according to the ISO C Std Annex
G. I found this out after porting some FreeBSD lib msun tests. Many edge
cases for complex floating point
Hi,
while running regression tests on an octeon ERL, I noticed that
no-macro-redefined flag was not known to the compiler.
I added some #undefs and some hours ago a change to art_walk made
the tests fail too.
The no-macro-redefine was added 3 month ago.
Thanks,
mbuhl
Index:
Hi,
I noticed that the iked regress test fails if I have an obj directory.
The following patch adresses this.
mbuhl
Index: regress/sbin/iked/parser/Makefile
===
RCS file: /mount/openbsd/cvs//src/regress/sbin/iked/parser/Makefile,v
not crash the program with other flags that use getpw{nam,uid}.
Thanks,
Moritz Buhl
Index: user.c
===
RCS file: /cvs/src/usr.sbin/user/user.c,v
retrieving revision 1.122
diff -u -p -r1.122 user.c
--- user.c 26 Sep 2018 14:54:58 -
and disabling/reenabling the
cores with hw.smt also works.
The only problem is that the lines reserved for cpu-stats does not
change with reenabling. Refreshing the output with '1' or resizing the
window should help.
Thanks,
Moritz Buhl
Index: display.c
Hi,
when rebooting a PowerBook G4, the kernel sometimes paniced while
detaching the disk.
Additionallly, during boot it was noted that / is not clean.
The following patch special cases polling to not return early anymore
and adds polling to all previous elements in the queue.
Please tell me if
Hello,
Removing my usb network interface during `tcpdump -i rum0` caused my
kernel to crash. A fix is attached.
Thanks,
Moritz Buhl
Index: net/bpf.c
===
RCS file: /cvs/src/sys/net/bpf.c,v
retrieving revision 1.169
diff -u -p
Sorry, I noticed that vgrind still is in the ports after writing the mail.
I will do more research before writing a patch next time.
On 11/08/16 11:49, Theo Buehler wrote:
On Tue, Nov 08, 2016 at 11:42:17AM +0100, Theo Buehler wrote:
On Tue, Nov 08, 2016 at 10:53:37AM +0100, Moritz Buhl wrote
Hi, since vgrind was removed in 4.9 I removed the -v option in ctags.
At least the example in ctags(1) could be modified.
Greetings Moritz Buhl
? ctags-vgrind.diff
Index: ctags.1
===
RCS file: /cvs/src/usr.bin/ctags/ctags.1,v
80 matches
Mail list logo
