Konstantin Ryabitsev wrote:
> On Tue, Mar 23, 2021 at 09:30:33AM -0700, James Bottomley wrote:
> > > I think the bulk of user issues are going to be regressions. Although
> > > you may be in a better position to know for sure, but at least for
> > > me, wearing my "user" hat, the thing that gets m
James Bottomley wrote:
> On Mon, 2021-03-22 at 13:16 -0400, Konstantin Ryabitsev wrote:
> > On Mon, Mar 22, 2021 at 04:18:14PM +0100, Thorsten Leemhuis wrote:
> > > Note, there is a second reason why ksummit-discuss is CCed: another
> > > reason why I want to create this new list is making it easi
Heiher wrote:
> Hello,
>
> I've added a pipe file descriptor (fd1) to an epoll (fd3) with
> EPOLLOUT in edge-triggered mode, and then added the fd3 to another
> epoll (fd4) with EPOLLIN in edge-triggered too.
>
> Next, waiting for fd4 without timeout. When fd1 to be writable, i
> think epoll_wai
Andrew Morton wrote:
> On Tue, 11 Jun 2019 16:54:44 +0200 Roman Penyaev wrote:
> > In order to measure polling from userspace libevent was modified [1] and
> > bench_http benchmark (client and server) was used:
> >
> > o EPOLLET, original epoll:
> >
> > 2 requests in 0.551306 sec. (362
Roman Penyaev wrote:
> epoll_create2() is needed to accept EPOLL_USERPOLL flags
> and size, i.e. this patch wires up polling from userspace.
Instead of adding a new syscall, is setting size (and/or even
the EPOLL_USEREPOLL flag) something that could be done via
ioctl?
There's no race like CLOEXE
Roman Penyaev wrote:
> Hi all,
+cc Jason Baron
> ** Limitations
> 4. No support for EPOLLEXCLUSIVE
> If device does not pass pollflags to wake_up() there is no way to
> call poll() from the context under spinlock, thus special work is
> scheduled to offload polling. In this s
Linus Torvalds wrote:
> On Tue, Jun 4, 2019 at 6:41 AM Oleg Nesterov wrote:
> >
> > This is the minimal fix for stable, I'll send cleanups later.
>
> Ugh. I htink this is correct, but I wish we had a better and more
> intuitive interface.
I had the same thoughts, but am not a regular kernel hac
cked by set_user_sigmask() can
> be delivered even if the caller returns success or timeout.
>
> Change restore_user_sigmask() to accept the additional "interrupted"
> argument which should be used instead of signal_pending() check, and
> u
"Eric W. Biederman" wrote:
> Frankly the only reason this appears to be worth touching is that we
> have a userspace regression. Otherwise I would call the current
> behavior more correct and desirable than ignoring the signal longer.
>
> If I am reading sitaution properly I suggest we go back t
David Laight wrote:
> From: Oleg Nesterov
> > Sent: 29 May 2019 17:12
> > Al, Linus, Eric, please help.
> >
> > The previous discussion was very confusing, we simply can not understand
> > each
> > other.
> >
> > To me everything looks very simple and clear, but perhaps I missed something
> > o
Andrew Morton wrote:
> On Tue, 21 May 2019 09:25:51 +0000 Eric Wong wrote:
>
> > Deepa Dinamani wrote:
> > > For all the syscalls that receive a sigmask from the userland,
> > > the user sigmask is to be in effect through the syscall execution.
> > >
or. If there is a signal and an error
> like EINVAL, the syscalls return -EINVAL rather than the interrupted
> error codes.
>
> The sys_io_uring_enter() seems to be returning success when there is
> a signal and the queue is not empty. This seems to be a bug.
Roman Penyaev wrote:
> diff --git a/fs/eventpoll.c b/fs/eventpoll.c
> index 81da4571f1e0..9d3905c0afbf 100644
> --- a/fs/eventpoll.c
> +++ b/fs/eventpoll.c
> @@ -44,6 +44,7 @@
> #include
> #include
> #include
> +#include
> #include
>
> /*
> @@ -185,6 +186,9 @@ struct epitem {
>
>
Deepa Dinamani wrote:
> Sorry, I was trying a new setup at work. I should have tested it.
> My bad, I've checked this one.
Thanks. This is good w.r.t. epoll_pwait and ppoll when applied
to 5.0.11 (no fs/io_uring.c).
Can't think of anything which uses pselect or aio on my system;
but it looks ri
Deepa Dinamani wrote:
> Eric,
> Can you please help test this?
Nope, that was _really_ badly whitespace-damaged.
(C'mon, it's not like you're new to this)
Deepa Dinamani wrote:
> So here is my analysis:
> So the 854a6ed56839a40f6 seems to be better than the original code in
> that it detects the signal.
OTOH, does matter to anybody that a signal is detected slightly
sooner than it would've been, otherwise?
> But, the problem is that it doesn't
Eric Wong wrote:
> (didn't test AIO, but everything else seems good)
"seems" != "is"
Now that I understand the fix for epoll, the fs/select.c changes
would hit the same problem and not return -EINTR when it should.
I'll let you guys decide how to fix this, but
Eric Wong wrote:
> Deepa Dinamani wrote:
> > I'm not sure what the hang in the userspace is about. Is it because
> > the syscall did not return an error or the particular signal was
> > blocked etc.
>
> Uh, ok; that's less comforting.
Nevermind,
Deepa Dinamani wrote:
> I was also not able to reproduce this.
> Arnd and I were talking about this today morning. Here is something
> Arnd noticed:
>
> If there was a signal after do_epoll_wait(), we never were not
> entering the if (err = -EINTR) at all before.
I'm not sure which `if' statemen
Davidlohr Bueso wrote:
> On Sun, 28 Apr 2019, Eric Wong wrote:
>
> > Just running one test won't trigger since it needs a busy
> > machine; but:
> >
> > make test/mgmt_auto_adjust.log
> > (and "rm make test/mgmt_auto_adjust.log" if you
Deepa Dinamani wrote:
> I tried to replicate the failure on qemu.
> I do not see the failure with N=32.
> Does it work for N < 32?
Depends on number of cores you have; I have 4 cores, 8 threads
with HT; so I needed to have a lot of load on the machine to get
it to fail (it takes about 1 minute).
Eric Wong wrote:
> Omar Kilani wrote:
> > Hi there,
> >
> > I’m still trying to piece together a reproducible test that triggers
> > this, but I wanted to post in case someone goes “hmmm... change X
> > might have done this”.
>
> Maybe Davidlohr knows,
Omar Kilani wrote:
> Hi there,
>
> I’m still trying to piece together a reproducible test that triggers
> this, but I wanted to post in case someone goes “hmmm... change X
> might have done this”.
Maybe Davidlohr knows, since he's responsible for most of the
epoll changes in 5.0.
> Basically, s
Bjorn Helgaas wrote:
> On Tue, Mar 5, 2019 at 5:26 PM Eric Wong wrote:
> > Bjorn Helgaas wrote:
>
> > > Any pointers? I guess there's no mutt backend that can read a
> > > public-inbox archive directly?
> >
> > There's mutt pat
Bjorn Helgaas wrote:
> OK, so I understand how to clone archives from lore.kernel.org and how
> to convert a git archive to a maildir (thanks, Konstantin!)
>
> What I *don't* understand is how to effectively read this locally.
> Ideally I'd like to run mutt, possibly with notmuch for indexing. B
Jakub Bogusz wrote:
> Let's try once again... (take 3)
>
> First time I sent this patch when I prepared it in 2013 for Linux 3.12,
> the second time after update for Linux 4.12 in 2017...
> And it still applies to Linux 4.20.
Adding Dmitry Torokhov and linux-input to Cc:
(that's what ./scripts/g
Joonas Lahtinen wrote:
> Quoting Eric Wong (2019-01-04 03:06:26)
> > Yeah, so the Debian bpo 4.17(.17) kernel did not set
> > CONFIG_INTEL_IOMMU_DEFAULT_ON, so I didn't encounter problems.
> > My self-built kernels all set CONFIG_INTEL_IOMMU_DEFAULT_ON.
>
> S
Alexandre Belloni wrote:
> Does this mean that it is set to an invalid value and that in is not
> getting updated properly in mc146818_set_time?
It seems to be set/updated without trouble while the machine is
on. It's only on a cold boot (from poweroff or hibernate) that
the value is invalid.
P
alues.
Fixes: 3c217e51d8a272b9 ("rtc: cmos: century support")
Cc: Alexandre Belloni
Cc: Alessandro Zummo
Cc: Sylvain Chouleur
Cc: Patrick McDermott
Cc: linux-...@vger.kernel.org
Signed-off-by: Eric Wong
---
drivers/rtc/rtc-mc146818-lib.c | 2 +-
1 file changed, 1 insertion(+), 1 de
Joey Pabalinas wrote:
> My only comment on the public-mailbox choice is that the documentation
> is very sparse and erratic. Myself and a couple other people just
> couldn't figure out how to convert that format to Maildir or some other
> format you could feed into a reader like neomutt.
Sorry, I
Joonas Lahtinen wrote:
> Quoting Eric Wong (2018-12-27 13:49:48)
> > I just got a used Thinkpad X201 (Core i5 M 520, Intel QM57
> > chipset) and hit some kernel panics while trying to view
> > image/animation-intensive stuff in Firefox (X11) unless I use
> > "iomm
I just got a used Thinkpad X201 (Core i5 M 520, Intel QM57
chipset) and hit some kernel panics while trying to view
image/animation-intensive stuff in Firefox (X11) unless I use
"iommu_intel=igfx_off".
With Debian stable backport kernels, "linux-image-4.17.0-0.bpo.3-amd64"
(4.17.17-1~bpo9+1) has n
Roman Penyaev wrote:
> On 2018-12-06 00:46, Eric Wong wrote:
> > Roman Penyaev wrote:
> > > Hi all,
> > >
> > > The goal of this patch is to reduce contention of ep_poll_callback()
> > > which
> > > can be called concurrently from differen
Roman Penyaev wrote:
> Hi all,
>
> The goal of this patch is to reduce contention of ep_poll_callback() which
> can be called concurrently from different CPUs in case of high events
> rates and many fds per epoll. Problem can be very well reproduced by
> generating events (write to pipe or event
_L
keycode 115 = Control_L NoSymbol Control_L
keycode 151 = Escape NoSymbol Escape
add Control = Control_L
Signed-off-by: Eric Wong
---
Documentation/laptops/thinkpad-acpi.txt | 11 +
drivers/platform/x86/thinkpad_acpi.c| 79 ++---
2 fi
entfd_poll made by Paolo.
Signed-off-by: Eric Wong
Cc: Paolo Bonzini
---
fs/pipe.c | 32 ++--
1 file changed, 30 insertions(+), 2 deletions(-)
diff --git a/fs/pipe.c b/fs/pipe.c
index 39d6f431da83..1a904d941cf1 100644
--- a/fs/pipe.c
+++ b/fs/pipe.c
@@ -509,7 +50
Ævar Arnfjörð Bjarmason wrote:
> Good point. I also see that (via git log --author=Ævar --grep='^\[PATCH
> ') that this series itself arrived out of order (0 -> 2 -> 1), but I
> don't know to what extent public-inbox itself might be batching things.
public-inbox doesn't batch, aside from when the
Hi Marek, I'm replying to
http://blog.cloudflare.com/the-sad-state-of-linux-socket-balancing/
via email so Jason and linux-kernel see it. I also don't believe in
using centralized, proprietary messaging like Disqus for discussing
Open Source, nor do I deal with JavaScript.
I still believe the bes
Marc Herbert wrote:
> PS: I used NNTP and http://dir.gmane.org/gmane.comp.version-control.git
> to quickly find this old thread (what could we do without NNTP?). Then
> I googled for a web archive of this thread and Google could only find
> this one:
> http://git.661346.n2.nabble.com/BUG-in-git-d
+Cc folks who may know about timer stuff on epoll.
Dmitry Banschikov wrote:
> Hi!
>
> I have a problem caused by inaccurate timeouts in epoll_wait(2).
> Here are some parts of strace -tt output:
Which kernel version are you using?
> 22578 09:33:46.959791 epoll_wait(5,
> 22578 09:33:50.010794
Sorry for the late response, forwarding to Dmitry and linux-input.
I haven't dealt with input in years...
Michael Shell wrote:
>
> Not a big deal, but something nonetheless ...
>
> I recently upgraded my kernel from 4.3 to 4.8.1 (this is a Linux From
> Scratch build). There was only one obvious
Zhouyi Zhou wrote:
> Hi,
> Yesterday, I cced 5 patches to linux-kernel@vger.kernel.org using
> send-email, but all of them were
> rejected by mail server at vger.kernel.org as follows:
> "Dear yizhouz...@ict.ac.cn, Your message to linux-kernel@vger.kernel.org
> was rejected by the recipient
Josh Triplett wrote:
> On August 9, 2016 11:37:31 PM HST, Richard Ipsum
> wrote:
>
> >Maybe there's a better solution to this problem than git-candidate
> >then,
> >maybe we can just invent some wonderful new subcommand that fetches
> >a mailing list archive into a git repo, for those that want
Josh Triplett wrote:
> On Mon, Aug 01, 2016 at 07:55:54AM +0000, Eric Wong wrote:
> > Christian Couder wrote:
> > > On Fri, Jul 29, 2016 at 12:10 PM, Richard Ipsum
> > > wrote:
> > > > On Thu, Jul 28, 2016 at 11:40:55PM -0700, Josh Triplett wrote:
>
Christian Couder wrote:
> On Fri, Jul 29, 2016 at 12:10 PM, Richard Ipsum
> wrote:
> > On Thu, Jul 28, 2016 at 11:40:55PM -0700, Josh Triplett wrote:
> > [snip]
> >>
> >> I'd welcome any feedback, whether on the interface and workflow, the
> >> internals and collaboration, ideas on presenting dif
+cc Jason Baron since he might be able to provide more insight into
epoll.
Mathias Krause wrote:
> Hi,
>
> this is an attempt to resurrect the thread initially started here:
>
> http://thread.gmane.org/gmane.linux.network/353003
>
> As that patch fixed the issue for the mentioned reproducer,
Madars Vitolins wrote:
> Hi Folks,
>
> I am developing kind of open systems application, which uses
> multiple processes/executables where each of them monitors some set
> of resources (in this case POSIX Queues) via epoll interface. For
> example when 10 processes on same queue are in state of e
-off-by: Eric Wong
Cc: Joe Turner
Cc: Takashi Iwai
---
sound/usb/quirks.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/sound/usb/quirks.c b/sound/usb/quirks.c
index 32631a8..2ea5b33 100644
--- a/sound/usb/quirks.c
+++ b/sound/usb/quirks.c
@@ -1118,6 +1118,7 @@ bool
SOCK_NONBLOCK, this does not affect the accepted socket, nor does
it change the file status flag of the listen socket for other calls.
Signed-off-by: Eric Wong
---
RFC since this seems a bit esoteric, and I'm not sure if it'd be
useful to others. I've certainly wished I've had it
Eric Wong wrote:
> I also had another generic patch prepared before I noticed Joe's
> quirk addition for the MS Lifecam HD-5000.
This is the generic patch I prepared before I noticed Joe's quirk
addition (applies cleanly to 3.19 and 4.0-rc6):
---
Pre" variant.
Signed-off-by: Eric Wong
Cc: Joe Turner
Cc: Takashi Iwai
---
I also had another generic patch prepared before I noticed Joe's
quirk addition for the MS Lifecam HD-5000.
sound/usb/quirks.c | 9 +++--
1 file changed, 7 insertions(+), 2 deletions(-)
diff --g
Jason Baron wrote:
> On 02/18/2015 12:51 PM, Ingo Molnar wrote:
> > * Ingo Molnar wrote:
> >
> >>> [...] However, I think the userspace API change is less
> >>> clear since epoll_wait() doesn't currently have an
> >>> 'input' events argument as epoll_ctl() does.
> >> ... but the change would be
Ingo Molnar wrote:
>
> * Ingo Molnar wrote:
>
> > > [...] However, I think the userspace API change is less
> > > clear since epoll_wait() doesn't currently have an
> > > 'input' events argument as epoll_ctl() does.
> >
> > ... but the change would be a bit clearer and somewhat
> > more fle
Jason Baron wrote:
> On 02/09/2015 11:49 PM, Eric Wong wrote:
> > Do you have a userland use case to share?
>
> I've been trying to describe the use case, maybe I haven't been doing a good
> job :(
Sorry, I meant if you had any public code.
Anyways, I've restar
Jason Baron wrote:
> On 02/09/2015 05:45 PM, Andy Lutomirski wrote:
> > On Mon, Feb 9, 2015 at 1:32 PM, Jason Baron wrote:
> >> On 02/09/2015 03:18 PM, Andy Lutomirski wrote:
> >>> On 02/09/2015 12:06 PM, Jason Baron wrote:
> Epoll file descriptors that are added to a shared wakeup source ar
Jason Baron wrote:
> I've done a bit of performance evaluation on a dual socket, 10 core, hyper
> threading enabled box: Intel(R) Xeon(R) CPU E5-2650 v3 @ 2.30GHz. For the
> simple epfdN->epfdN->pipefdN topology case where each thread has its
> own unique files and is doing EPOLL_CTL_ADD and EPOLL
quot;epoll_ctl(op=EPOLL_CTL_DEL)")
>
> Remove use of epds in epoll_ctl when op == EPOLL_CTL_DEL.
>
> Fixes: 4d7e30d98939 ("epoll: Add a flag, EPOLLWAKEUP, to prevent suspend
> while epoll events are ready")
> Signed-off-by: Nicolas Iooss
Looks good t
Steven Stewart-Gallus wrote:
> I'm trying to debug a hangup where my program loops with FUTEX_WAIT (actually
> FUTEX_WAIT_PRIVATE but same thing) endlessly erring out with EAGAIN. I would
> like to know if anyone on the mailing list knows when FUTEX_WAIT can fail with
> EAGAIN.
FUTEX_WAIT fails
Sasha Levin wrote:
> On 05/15/2014 02:11 PM, Peter Zijlstra wrote:
> > On Mon, May 12, 2014 at 11:42:33AM -0400, Sasha Levin wrote:
> >> Hi all,
> >>
> >> While fuzzing with trinity inside a KVM tools guest running the latest
> >> -next
> >> kernel I've stumbled on the following spew. Maybe relat
Nathaniel Yazdani wrote:
> On Sun, Feb 23, 2014 at 9:32 PM, Eric Wong wrote:
> > Nathaniel Yazdani wrote:
> >> +asmlinkage long sys_epoll(int ep, struct epoll __user *in,
> >> + unsigned int inc, struct epoll __user *out,
> >> +
Nathaniel Yazdani wrote:
> + * stores triggered eventpoll entries in the 'out' array. The input array is
> + * _not_ read-only, because the resulting event mask gets written back to
> each
> + * entry's ->ep_events field. When successful, this will be the same as
> before
> + * (plus EPOLLERR &
Nathaniel Yazdani wrote:
> +asmlinkage long sys_epoll(int ep, struct epoll __user *in,
> + unsigned int inc, struct epoll __user *out,
> + unsigned int outc, int timeout);
I can understand using the new struct for 'in', but 'out' could just be
"struct e
Nathaniel Yazdani wrote:
> +++ b/include/uapi/linux/eventpoll.h
> @@ -61,6 +61,12 @@ struct epoll_event {
> __u64 data;
> } EPOLL_PACKED;
>
> +struct epoll {
> + int ep_fildes; /* file descriptor */
> + int ep_events; /* triggering events */
> + long long ep_ident; /* entry ID
Andy Lutomirski wrote:
> >> On 02/02/2014 06:17 PM, Nathaniel Yazdani wrote:
> > So are you saying that those features you mentioned are specifically sought
> > after for the kernel? If so I'd like to take a crack at some of them,
> > may as well
> > get some use out of my new knowledge of epoll i
Oleg Nesterov wrote:
> Yes. Before that 971316f0503a hack epoll can't even know if the task
> which did signalfd_poll() exits and frees the active signalfd_wqh.
> If for example that task forked a child before exit.
>
> And the whole RCU logic is only needed if exit/ep_remove_wait_queue
> actuall
Andi Kleen wrote:
> From: Andi Kleen
>
> A static checker was pointing out that nothing can possible set
> nwait < 0 in this path. The comment and the check appears to be
> outdated. Remove it.
I don't think so...
> Cc: v...@zeniv.linux.org.uk
> Signed-off-by: Andi Kleen
> ---
> fs/eventpoll
Jason Baron wrote:
> epoll: reduce usage of global 'epmutex' lock
>
> Epoll file descriptors that are 1 link from a wakeup source and
> are not nested within other epoll descriptors, or pointing to
> other epoll descriptors, don't need to check for loop creation or
> the creati
Zach Brown wrote:
> Towards the end of that thread Eric Wong asked why we didn't just
> extend splice. I immediately replied with some dumb dismissive
> answer. Once I sat down and looked at it, though, it does make a
> lot of sense. So good job, Eric. +10 Dummie points fo
Mayk Eskila wrote:
> Hello list,
>
> I intend to upgrade my own disc and partition cloning program by
> using splice. When running splice-cp.c from current git on my
> average dualcore hardware with kernel 3.10 I found that copying 1G
> files with splice is about twice as fast as using plain vani
dormando wrote:
> > I noticed these warnings on stock 3.10.9 running stress tests on
> > cmogstored.git (git://bogomips.org/cmogstored.git) doing standard
> > HTTP server stuff between lo and tmpfs:
> >
> [...]
> > I was going to reboot into 3.10.10 before I looked at dmesg. These
> > warnings ha
Eric Wong wrote:
> I noticed these warnings on stock 3.10.9 running stress tests on
> cmogstored.git (git://bogomips.org/cmogstored.git) doing standard
> HTTP server stuff between lo and tmpfs:
I'm still running the same test (in-place upgraded cmogstored a few
times to fix s
I noticed these warnings on stock 3.10.9 running stress tests on
cmogstored.git (git://bogomips.org/cmogstored.git) doing standard
HTTP server stuff between lo and tmpfs:
HTTP PUT requests hit: read(socket) + pwrite()
HTTP GET requests hit: send(MSG_MORE) + sendfile()
HTTP/1.1 persistent connecti
zet
> Cc: Al Viro
> Cc: Theodore Ts'o
Seems reasonable. This isn't a remotely hot path and f_op->release
runs in a workqueue nowadays (right?).
Acked-by: Eric Wong
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a messag
Eliezer Tamir wrote:
> I'm not sure I understand your claim that epoll has a higher latency
> than poll.
Nevermind, I usually live in an EPOLLONESHOT bubble so I use more
epoll_ctl than "normal" epoll apps :x
Anyways, good to know epoll wins over poll. Thanks.
--
To unsubscribe from this list:
Eliezer Tamir wrote:
> Performance:
> using sockperf, Intel X520 NICs,
> Supermicro 6026TT-BTF systems with E5-2690 Xeon CPUs
> 100 UDP sockets avg. latency 5.756 (std-dev 0.510)
> 1k UDP sockets avg. latency 5.780 (std-dev 0.536)
> 10k UDP sockets avg. latency 6.269 (std-dev 0.611)
How does thi
Eric Dumazet wrote:
> On Fri, 2013-07-19 at 23:50 +0000, Eric Wong wrote:
> > Eric Dumazet wrote:
> > > Hi Al
> > >
> > > I tried to debug strange crashes in tcp_poll() called from
> > > sys_epoll_wait() -> sock_poll()
> > >
&g
Eric Dumazet wrote:
> Hi Al
>
> I tried to debug strange crashes in tcp_poll() called from
> sys_epoll_wait() -> sock_poll()
>
> The symptom is that sock->sk is NULL and we therefore dereference a NULL
> pointer.
>
> It's really rare crashes but still, it would be nice to understand where
> is
Zach Brown wrote:
> On Wed, May 15, 2013 at 07:44:05PM +0000, Eric Wong wrote:
> > Why introduce a new syscall instead of extending sys_splice?
>
> Personally, I think it's ugly to have different operations use the same
> syscall just because their arguments match.
Fair
Zach Brown wrote:
> This adds a syscall and vfs entry point for clone_range which offloads
> data copying between existing files.
>
> The syscall is a thin wrapper around the vfs entry point. Its arguments
> are inspired by sys_splice().
Why introduce a new syscall instead of extending sys_spli
Cc-ing Jens
David Oostdyk wrote:
> Hello,
>
> I have a few relatively high-end systems with hardware RAIDs which
> are being used for recording systems, and I'm trying to get a better
> understanding of contiguous write performance.
>
> The hardware that I've tested with includes two high-end I
Cc:-ing netdev (no comments of my own)
Felix Becker wrote:
> Hi,
>
> I tried to figure out how old my TCP connections are and took a look
> at /proc//fd/ using 'ls -la' / 'stat'.
>
> When I'm creating a new socket in my application, the time stamps
> returned by stat / ls -la are correct - as e
Mathieu Desnoyers wrote:
> It is provided as a starting point only. Test cases should be ported
> from Userspace RCU to kernel space and thoroughly ran on a wide range of
> architectures before considering this port production-ready.
Hi Mathieu, any progress on this front? I'm not sure how to ap
me ./eponeshotmt -c 100 -w 4 -t 4 -f 10
real0m 5.83s
user0m 1.35s
sys 0m 21.95s
I also ran v2 on Davide Libenzi's totalmess epoll stresser for a few
hours yesterday without failures. Running totalmess right now on v3,
so far so good :)
Tested-by: Eric Wong
--
To unsubscribe from
Neil Horman wrote:
> On Fri, Apr 05, 2013 at 07:38:55PM +0000, Eric Wong wrote:
> > Hi Neil, I'm wondering if you have or know of any public
> > documentation/examples for using net_dropmon.
> >
> > If not, I'll figure it out on my own at some point.
>
as worried about the cmpxchg at first, but it does not seem to hurt
performance on my 4-core system. In fact, it was slightly better
(but within margin of error)
time ./eponeshotmt -c 100 -w 4 -t 4 -f 10
real 0m 5.78s
user0m 1.20s
sys 0m 21.90s
Tested-by: Eric Wong
Hopefully somebody can
Hi Neil, I'm wondering if you have or know of any public
documentation/examples for using net_dropmon.
If not, I'll figure it out on my own at some point.
Thanks in advance!
(Not a very high priority project for me, my network connectivity
problems are sadly _very_ obvious at the moment :x)
--
Mathieu Desnoyers wrote:
> * Eric Wong (normalper...@yhbt.net) wrote:
> > In some situations, it is necessary to prepend a node to a queue.
> > For epoll, this is necessary for two rare conditions:
> >
> > * when the user triggers -EFAULT
> > * when rein
machines with 64-byte cache line sizes (or less), this
saves us from accessing the extra cache line and helps improve performance.
* destroy wakeup source immediately for zombie epitems, there is no need
to wait until ep_send_events.
Signed-off-by: Eric Wong
Cc: Davide Libenzi
Cc: Al Viro
C
Pavel Machek wrote:
> Eric Wong wrote:
> > [1] my splice() annoyances:
> > * need to create/manage a pipe
> > * copy size limited by pipe size
> > * doesn't reduce userspace syscalls (just data copy overhead)
> > * easy to misuse and star
In some situations, it is necessary to prepend a node to a queue.
For epoll, this is necessary for two rare conditions:
* when the user triggers -EFAULT
* when reinjecting elements from the ovflist (implemented as a stack)
Signed-off-by: Eric Wong
Cc: Mathieu Desnoyers
---
This is on top of
It is always safe to use RCU_INIT_POINTER to NULL a pointer.
This results in slightly smaller/faster code.
Signed-off-by: Eric Wong
Cc: Andrew Morton
---
Andrew: Sorry for the noise and requiring these cleanups. Would you
want a squashed commit with all my RCU+ws-related epoll changes to
Oleg Nesterov wrote:
> On 03/14, Eric Wong wrote:
> > Oleg Nesterov wrote:
> > > On 03/10, Eric Wong wrote:
> > > >
> > > > This fixes the following sparse error when using
> > > > CONFIG_SPARSE_RCU_POINTER=y and "make C=2 fs/
Fredrik Tolf wrote:
> It is worth noting, also, that this seems to be a situation
> introduced somewhere between 2.6.26 and 2.6.32, because I started
> noticing it when I upgraded from Debian 5.0 to 6.0. I've since tried
> it on 3.2.0, 3.5.4 and 3.7.1, and it appears in every version.
> However, I
Mathieu Desnoyers wrote:
> * Eric Wong (normalper...@yhbt.net) wrote:
> > /*
> > + * ___wfcq_append: append one local queue to another local queue
> > + *
> __wfcq_append() and ___wfcq_append() are meant to be private to
> wfcqueue.h. Therefore, the comment above sho
-off-by: Eric Wong
Cc: Mathieu Desnoyers
Cc: Lai Jiangshan
Cc: Paul E. McKenney
Cc: Stephen Hemminger
Cc: Davide Libenzi
---
I noticed the original __wfcq_append function was not in the
synchronization table, so I left out ___wfcq_append from the table
as well.
include/linux/wfcqueue.h
Sergey Senozhatsky wrote:
> [3.163894] ===
> [3.163895] [ INFO: suspicious RCU usage. ]
> [3.163897] 3.9.0-rc3-next-20130322-dbg-dirty #1 Not tainted
> [3.163898] ---
> [3.163900] fs/eventpoll.c:545 suspicious rcu_derefere
Eric Wong wrote:
> Arve Hjønnevåg wrote:
> >
> > At some point the level triggered event has to get cleared. As far as
> > I can tell, your new code will drop new events that occur between
> > "revents = ep_item_poll(epi, &pt);" and "epi->stat
Arve Hjønnevåg wrote:
> On Fri, Mar 22, 2013 at 3:31 AM, Eric Wong wrote:
> > Arve Hjønnevåg wrote:
> >> On Thu, Mar 21, 2013 at 8:24 PM, Eric Wong wrote:
> >> >
> >> > With EPOLLET and improper usage (not hitting EAGAIN), the event now
> >> &
Eric Wong wrote:
> Arve Hjønnevåg wrote:
> > On Thu, Mar 21, 2013 at 8:24 PM, Eric Wong wrote:
> > >
> > > With EPOLLET and improper usage (not hitting EAGAIN), the event now
> > > has a larger window to be lost (as mentioned in my changelog).
> > >
Mathieu Desnoyers wrote:
> * Eric Wong (normalper...@yhbt.net) wrote:
> > /*
> > + * __wfcq_append_local: append one local queue to another local queue
> > + *
> > + * No memory barriers are issued. Mutual exclusion is the responsibility
> > + * of the caller.
1 - 100 of 200 matches
Mail list logo
