Alexander Bluhm wrote:
> Several possible solutions like letting tasks sleep at a different
> priority or changing kernel threads to user priority have been
> discussed. There was no consensus what to do.
>
> To move forward and uncouple the splicing improvement from the
> scheduler discussion, I
When in list mode, the filenames in the archive are written to stdout.
For pax, POSIX says that *only* the file listing output may be written
there; all diagnostic output must go to stderr. So don't use listf or
listfd to write non-file-listing output.
There's a tricky bit that I'm not sure
On Thu, Aug 25, 2016 at 12:49:45AM +0200, Mark Kettenis wrote:
> So enabling full hardware-enforced W^X is perhaps a bit to much at
> this moment. But enabling the bit that enforces that the kernel
> cannot execute pages that are writable by userland should be fine.
> The diff does this on all ARM
pax includes routines for caching getpw{nam,uid} and getgr{nam,gid}
results. Okay, but libc already has caching uid->name and gid->name
routines: user_from_uid() and group_from_gid(). Let's switch pax over to
the libc routines. That'll free up a little space on the ramdisks.
Notes:
1) the pa
So enabling full hardware-enforced W^X is perhaps a bit to much at
this moment. But enabling the bit that enforces that the kernel
cannot execute pages that are writable by userland should be fine.
The diff does this on all ARMv7 processors that include the
Virtualization Extensions, which is what
On Fri, Jul 29, 2016 at 08:07:14PM -0400, Ted Unangst wrote:
> Alexander Bluhm wrote:
> > On Fri, Jul 29, 2016 at 06:46:52PM -0400, Ted Unangst wrote:
> > > Alexander Bluhm wrote:
> > > > + /* Avoid user land starvation. */
> > > > + yield();
> > >
> > > you don't need to yield here, t
On Wed, Aug 24, 2016 at 05:52:39PM +0200, Martin Pieuchot wrote:
> Diff below simplifies ip6_getpmtu() to use a 'struct rtentry *' instead
> of two 'struct route_in6'.
>
> ok?
I have tested it with regress/sys/netinet/pmtu .
OK bluhm@
> + if (rt != NULL) {
...
> - } else if (ifp0) {
> -
On 24.08.2016 22:03, Ingo Schwarze wrote:
>> Could you show some code? In my testing fgetwln() fails on next read if
>> previously there was partial line with tail EILSEQ. Stdio not advance
>> its pointer over the sequence with EILSEQ.
>
> See below for a radically stripped down version of FreeBSD
per freebsd bug report:
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=201650
echo "1234 1234 1234" | grep -o ^
1234
Index: util.c
===
RCS file: /cvs/src/usr.bin/grep/util.c,v
retrieving revision 1.55
diff -u -p -r1.55 util.c
Hi Andrey,
Andrey Chernov wrote on Wed, Aug 24, 2016 at 09:15:15PM +0300:
> And if we plan to change original 44lite
> function behavior, all BSD camps should agree at least.
I agree that it is preferable to do such bugfixes in consensus, if
possible; that helps to improve quality and coherence.
On 24.08.2016 20:49, Ingo Schwarze wrote:
>> Yes, it is a bit problematic, but it is the way this function
>> designed in BSD 44lite, so I disagree with the patch.
>> I.e. preserving partial line is more essential for it than
>> returning no line at all (NULL), so tail error on partial line
>> is n
On Wed, 24 Aug 2016 19:49:47 +0200, Ingo Schwarze wrote:
> See the improved patch below!
Looks good. I wouldn't bother with an else after the break but if
you think it reads better that way, that's fine with me.
- todd
Hi Todd, hi Andrey,
Todd C. Miller wrote on Wed, Aug 24, 2016 at 10:03:47AM -0600:
> I'm probably not understanding something here
No, i think yours is a great idea, i merely failed to think of it.
> but why can't you just test for __SEOF in fp->_flags
> when __srefill() returns non-zero
> and
On 24.08.2016 18:33, Ingo Schwarze wrote:
> 2. The stdio read buffer is not empty at the time of the call
> or the first __refill() succeeds, but then a subsequent __refill()
> fails:
> fgetln(3) returns a buffer containing the concatenation of all
> strings that were returned by t
I'm probably not understanding something here but why can't you
just test for __SEOF in fp->_flags when __srefill() returns non-zero
and treat the absence of __SEOF as an error?
- todd
Diff below simplifies ip6_getpmtu() to use a 'struct rtentry *' instead
of two 'struct route_in6'.
ok?
Index: netinet6/ip6_output.c
===
RCS file: /cvs/src/sys/netinet6/ip6_output.c,v
retrieving revision 1.212
diff -u -p -r1.212 ip6_o
Hi,
what happens in fgetln(3) when the buffer runs out and reading more
bytes fails before a newline character is found? There are two
cases:
1. The stdio read buffer is empty at the time of the call,
and the read(2) in __refill() fails:
fgetln(3) returns NULL and sets errno and the std
Hi,
been running w/this on wandboard for some days building ports.
pm_cstate should go too, but it's another small diff.
-Artturi
diff --git a/sys/arch/arm/arm/cpuswitch7.S b/sys/arch/arm/arm/cpuswitch7.S
index 2eeecec..3196c8e 100644
--- a/sys/arch/arm/arm/cpuswitch7.S
+++ b/sys/arch/arm/arm/c
On Wed, Aug 24, 2016 at 11:51:56AM +0200, Mark Kettenis wrote:
> The pmap_fault_fixup() functions does a number of different fixups:
>
> 1. page modified emulation
>
> 2. page access emulation
>
> 3. entering level-1 entries for level-2 page tables
>
> Since it is a generic function that gets c
The pmap_fault_fixup() functions does a number of different fixups:
1. page modified emulation
2. page access emulation
3. entering level-1 entries for level-2 page tables
Since it is a generic function that gets called for many fault types,
it has to be careful to only enter mappings that are
i noticed this when trying to figure out what level a cardslot would
interrupt at.
now we have splraise this is a lot more straightforward.
avoiding an splraise when we're not sure whats actually lower than
IPL_BIO is not worth the effort. splraise to a lower level is fine.
ok?
Index: pccbb.c
=
21 matches
Mail list logo