On 05/25/2018 10:23 AM, David Miller wrote:
> From: Qing Huang
> Date: Wed, 23 May 2018 16:22:46 -0700
>
>> When a system is under memory presure (high usage with fragments),
>> the original 256KB ICM chunk allocations will likely trigger kernel
>> memory management to enter slow path doing me
On 05/29/2018 07:36 AM, Yafang Shao wrote:
> On Tue, May 29, 2018 at 10:15 PM, David Miller wrote:
>> From: Yafang Shao
>> Date: Fri, 25 May 2018 18:14:05 +0800
>>
>>> skb->len is meaningless to user.
>>> data length could be more helpful, with which we can easily filter out
>>> the packet wit
On 05/28/2018 05:41 PM, Yafang Shao wrote:
> OK.
>
> And what about introducing a new helper tcp_hdr_fast() ?
>
> /* use it when tcp header has not been pulled yet */
> static inline struct tcphdr *tcp_hdr_fast(const struct sk_buff *skb)
>
> {
>
> return (const struct tcphdr *)skb->
On Mon, May 28, 2018 at 8:36 AM Yafang Shao wrote:
> This is additional to the commit ea1627c20c34 ("tcp: minor optimizations
around tcp_hdr() usage").
> At this point, skb->data is same with tcp_hdr() as tcp header has not
> been pulled yet.
> Cc: Eric Dumazet
>
On 05/21/2018 11:09 AM, David Miller wrote:
> From: syzbot
> Date: Mon, 21 May 2018 11:05:02 -0700
>
>> find_match+0x244/0x13a0 net/ipv6/route.c:691
>> find_rr_leaf net/ipv6/route.c:729 [inline]
>> rt6_select net/ipv6/route.c:779 [inline]
>
> Hmmm, endless loop in find_rr_leaf or similar?
>
On 05/21/2018 12:55 AM, YueHaibing wrote:
> like commit df39a9f106d5 ("bpf: check NULL for sk_to_full_sk() return value"),
> we should check sk_to_full_sk return value against NULL.
>
> Signed-off-by: YueHaibing
> ---
> include/linux/bpf-cgroup.h | 2 +-
> 1 file changed, 1 insertion(+), 1 del
SCTP experts, please take a look.
On 05/19/2018 08:55 AM, syzbot wrote:
> Hello,
>
> syzbot found the following crash on:
>
> HEAD commit: 73fcb1a370c7 Merge branch 'akpm' (patches from Andrew)
> git tree: upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=1462ec0f8000
On 05/18/2018 05:08 AM, DaeRyong Jeong wrote:
> We report the crash: WARNING in ip_recv_error
> (I resend the email since I mistakenly missed the subject in my previous
> email. I'm sorry.)
>
>
> This crash has been found in v4.17-rc1 using RaceFuzzer (a modified
> version of Syzkaller), which
On 05/17/2018 01:53 PM, Qing Huang wrote:
> When a system is under memory presure (high usage with fragments),
> the original 256KB ICM chunk allocations will likely trigger kernel
> memory management to enter slow path doing memory compact/migration
> ops in order to complete high order memory a
On 05/15/2018 11:53 AM, Qing Huang wrote:
>
>> This is control path so it is less latency-sensitive.
>> Let's not produce unnecessary degradation here, please call kvzalloc so we
>> maintain a similar behavior when contiguous memory is available, and a
>> fallback for resiliency.
>
> No sure
On 05/13/2018 12:00 AM, Christophe JAILLET wrote:
> When 'kvzalloc()' is used to allocate memory, 'kvfree()' must be used to
> free it.
>
> Signed-off-by: Christophe JAILLET
> ---
> v1 -> v2: More places to update have been added to the patch
Please add relevant Fixes: tag(s)
On 05/11/2018 11:41 AM, Marcelo Ricardo Leitner wrote:
> But calling ip6_xmit with rcu_read_lock is expected. tcp stack also
> does it.
> Thus I think this is more of an issue with IPv6 stack. If a host has
> an extensive ip6tables ruleset, it probably generates this more
> easily.
>
>>> sctp_
On 05/11/2018 04:07 AM, Christoph Hellwig wrote:
> Signed-off-by: Christoph Hellwig
> ---
> include/net/tcp.h | 4 ++--
> net/ipv4/af_inet.c | 3 ++-
> net/ipv4/tcp.c | 31 ++-
> net/ipv6/af_inet6.c | 3 ++-
> 4 files changed, 20 insertions(+), 21 deletion
On 05/09/2018 11:53 PM, Jon Maxwell wrote:
> This version has some suggestions by Eric Dumazet:
>
> - Use a local variable for the mark in IPv6 instead of ctl_sk to avoid SMP
> races.
> - Use the more elegant "IP4_REPLY_MARK(net, skb->mark) ?: sk->sk_mark"
>
On 05/09/2018 10:21 PM, Jon Maxwell wrote:
...
> if (th->rst)
> @@ -723,11 +724,17 @@ static void tcp_v4_send_reset(const struct sock *sk,
> struct sk_buff *skb)
> arg.tos = ip_hdr(skb)->tos;
> arg.uid = sock_net_uid(net, sk && sk_fullsock(sk) ? sk : NULL);
> local_bh_d
On 05/09/2018 07:07 PM, Jon Maxwell wrote:
> Aidan McGurn from Openwave Mobility systems reported the following bug:
>
> "Marked routing is broken on customer deployment. Its effects are large
> increase in Uplink retransmissions caused by the client never receiving
> the final ACK to their FI
On 05/09/2018 12:21 PM, Willem de Bruijn wrote:
> Indeed. The skb shared info struct is zeroed by dev_validate_header
> as a result of dev->hard_header_len exceeding skb->end - skb->data.
>
> Not exactly sure yet how this can happen. The hard header length space
> is accounted for during alloca
On 05/07/2018 10:54 PM, Cong Wang wrote:
> On Mon, May 7, 2018 at 10:27 PM, syzbot
> wrote:
>> Hello,
>>
>> syzbot found the following crash on:
>>
>> HEAD commit:75bc37fefc44 Linux 4.17-rc4
>> git tree: upstream
>> console output: https://syzkaller.appspot.com/x/log.txt?x=1162c6978000
On 05/07/2018 07:16 PM, Jia-Ju Bai wrote:
> Yes, "&dev->stats" will not change, because it is a fixed address.
> But the field data in "dev->stats" is changed (rx_frame_errors, rx_crc_errors
> and rx_missed_errors).
> So if the driver returns "&dev->stats" without lock protection (like on line
On 05/07/2018 05:51 PM, Jia-Ju Bai wrote:
>
>
> On 2018/5/7 22:15, Eric Dumazet wrote:
>>
>> On 05/07/2018 07:08 AM, Jia-Ju Bai wrote:
>>> The write operations to "dev->stats" are protected by
>>> the spinlock on line 862-864, but the read
On 05/07/2018 07:08 AM, Jia-Ju Bai wrote:
> The write operations to "dev->stats" are protected by
> the spinlock on line 862-864, but the read operations to
> this data on line 858 and 867 are not protected by the spinlock.
> Thus, there may exist data races for "dev->stats".
>
> To fix the dat
On 05/05/2018 09:40 AM, Greg KH wrote:
> On Sat, May 05, 2018 at 08:47:02AM -0700, syzbot wrote:
>> Hello,
>>
>> syzbot found the following crash on:
>>
>> HEAD commit:8fb11a9a8d51 net/ipv6: rename rt6_next to fib6_next
>> git tree: net-next
>> console output: https://syzkaller.appspot.
On 05/04/2018 02:08 PM, Chintan Shah wrote:
> A UDP application which opens multiple sockets with same local
> address/port combination (using SO_REUSEPORT/SO_REUSEADDR socket options);
> and issues connect to a remote socket (using one of these local socket).
> Now if the same socket, which issu
number of bytes
that should be read using conventional read()/recv()/recvmsg() system calls,
to skip a sequence of bytes that can not be mapped, because not properly page
aligned.
Signed-off-by: Eric Dumazet
Cc: Andy Lutomirski
Acked-by: Soheil Hassas Yeganeh
---
tools/testing/selftests/net
int in case user request was completed.
Eric Dumazet (2):
tcp: add TCP_ZEROCOPY_RECEIVE support for zerocopy receive
selftests: net: tcp_mmap must use TCP_ZEROCOPY_RECEIVE
include/uapi/linux/tcp.h | 8 +
net/ipv4/af_inet.c | 2 +
net/ipv4
f mmap() and setsockopt(... TCP_ZEROCOPY_RECEIVE ...)
Note that memcg might require additional changes.
Fixes: 93ab6cc69162 ("tcp: implement mmap() for zero copy receive")
Signed-off-by: Eric Dumazet
Reported-by: syzbot
Suggested-by: Andy Lutomirski
Cc: linux...@kvack.org
Acked-by:
On Fri, Apr 27, 2018 at 1:45 AM kbuild test robot wrote:
> Hi Eric,
> Thank you for the patch! Yet something to improve:
> [auto build test ERROR on net-next/master]
> url:
https://github.com/0day-ci/linux/commits/Eric-Dumazet/tcp-add-TCP_ZEROCOPY_RECEIVE-support-for-zerocopy-rece
On 04/26/2018 02:16 PM, Andy Lutomirski wrote:
> At the risk of further muddying the waters, there's another minor tweak
> that could improve performance on certain workloads. Currently you mmap()
> a range for a given socket and then getsockopt() to receive. If you made
> it so you could mmap(
On 04/25/2018 06:20 PM, Soheil Hassas Yeganeh wrote:
>
> Acked-by: Soheil Hassas Yeganeh
>
>
Thanks Soheil for reviewing.
I have changed setsockopt() to getsockopt() so chose to not carry your Acked-by
Please add it back if you agree, thanks !
f mmap() and setsockopt(... TCP_ZEROCOPY_RECEIVE ...)
Note that memcg might require additional changes.
Fixes: 93ab6cc69162 ("tcp: implement mmap() for zero copy receive")
Signed-off-by: Eric Dumazet
Reported-by: syzbot
Suggested-by: Andy Lutomirski
Cc: linux...@kvack.org
Cc:
number of bytes
that should be read using conventional read()/recv()/recvmsg() system calls,
to skip a sequence of bytes that can not be mapped, because not properly page
aligned.
Signed-off-by: Eric Dumazet
Cc: Andy Lutomirski
Cc: Soheil Hassas Yeganeh
---
tools/testing/selftests/net/tcp_mmap.c
.
v3: change TCP_ZEROCOPY_RECEIVE to be a getsockopt() option
instead of setsockopt(), feedback from Ka-Cheon Poon
v2: Added a missing page align of zc->length in tcp_zerocopy_receive()
Properly clear zc->recv_skip_hint in case user request was completed.
Eric Dumazet (2):
tc
On 04/26/2018 06:40 AM, Ka-Cheong Poon wrote:
> A quick question. Is it a normal practice to return a result
> in setsockopt() given that the optval parameter is supposed to
> be a const void *?
Very good question.
Andy suggested an ioctl() or setsockopt(), and I chose setsockopt() but it loo
f mmap() and setsockopt(... TCP_ZEROCOPY_RECEIVE ...)
Note that memcg might require additional changes.
Fixes: 93ab6cc69162 ("tcp: implement mmap() for zero copy receive")
Signed-off-by: Eric Dumazet
Reported-by: syzbot
Suggested-by: Andy Lutomirski
Cc: linux...@kvack.org
Cc:
number of bytes
that should be read using conventional read()/recv()/recvmsg() system calls,
to skip a sequence of bytes that can not be mapped, because not properly page
aligned.
Signed-off-by: Eric Dumazet
Cc: Andy Lutomirski
Cc: Soheil Hassas Yeganeh
---
tools/testing/selftests/net/tcp_mmap.c
.
v2:
Added a missing page align of zc->length in tcp_zerocopy_receive()
Properly clear zc->recv_skip_hint in case user request was completed.
Eric Dumazet (2):
tcp: add TCP_ZEROCOPY_RECEIVE support for zerocopy receive
selftests: net: tcp_mmap must use TCP_ZEROCOPY_RECEIVE
include/uapi
On 04/25/2018 09:35 AM, Eric Dumazet wrote:
>
>
> On 04/25/2018 09:22 AM, Andy Lutomirski wrote:
>
>> In general, I suspect that the zerocopy receive mechanism will only
>> really be a win in single-threaded applications that consume large
>> amounts of rec
On 04/25/2018 09:22 AM, Andy Lutomirski wrote:
> In general, I suspect that the zerocopy receive mechanism will only
> really be a win in single-threaded applications that consume large
> amounts of receive bandwidth on a single TCP socket using lots of
> memory and don't do all that much else.
On 04/25/2018 09:04 AM, Matthew Wilcox wrote:
> If you don't zap the page range, any of the CPUs in the system where
> any thread in this task have ever run may have a TLB entry pointing to
> this page ... if the page is being recycled into the page allocator,
> then that page might end up as a
On 04/24/2018 10:27 PM, Eric Dumazet wrote:
> When adding tcp mmap() implementation, I forgot that socket lock
> had to be taken before current->mm->mmap_sem. syzbot eventually caught
> the bug.
> +
...
> + down_read(¤t->mm->mmap_sem);
> +
> + ret
On 04/24/2018 11:28 PM, Christoph Hellwig wrote:
> On Tue, Apr 24, 2018 at 10:27:21PM -0700, Eric Dumazet wrote:
>> When adding tcp mmap() implementation, I forgot that socket lock
>> had to be taken before current->mm->mmap_sem. syzbot eventually caught
>> the bug.
f mmap() and setsockopt(... TCP_ZEROCOPY_RECEIVE ...)
Note that memcg might require additional changes.
Fixes: 93ab6cc69162 ("tcp: implement mmap() for zero copy receive")
Signed-off-by: Eric Dumazet
Reported-by: syzbot
Suggested-by: Andy Lutomirski
Cc: linux...@kvack.org
Cc:
.
Eric Dumazet (2):
tcp: add TCP_ZEROCOPY_RECEIVE support for zerocopy receive
selftests: net: tcp_mmap must use TCP_ZEROCOPY_RECEIVE
include/uapi/linux/tcp.h | 8 ++
net/ipv4/tcp.c | 186 +
tools/testing/selftests/net/tcp_mmap.c
number of bytes
that should be read using conventional read()/recv()/recvmsg() system calls,
to skip a sequence of bytes that can not be mapped, because not properly page
aligned.
Signed-off-by: Eric Dumazet
Cc: Andy Lutomirski
Cc: Soheil Hassas Yeganeh
---
tools/testing/selftests/net/tcp_mmap.c
On 04/24/2018 04:47 AM, Yafang Shao wrote:
>
> Could you pls. explain the issue to me ?
Just run a synflood test on your host, it will definitely show the atomic
consuming most cpu cycles in inet_reqsk_alloc(), because of huge contention
on a cache line shared by all cpus.
Performance is redu
On 04/23/2018 09:39 PM, Yafang Shao wrote:
> On Tue, Apr 24, 2018 at 12:09 AM, Eric Dumazet wrote:
>>
>>
>> On 04/23/2018 08:58 AM, David Miller wrote:
>>> From: Yafang Shao
>>> Date: Sun, 22 Apr 2018 21:50:04 +0800
>>>
>>>> With
On 04/23/2018 07:04 PM, Andy Lutomirski wrote:
> On Mon, Apr 23, 2018 at 2:38 PM, Eric Dumazet wrote:
>> Hi Andy
>>
>> On 04/23/2018 02:14 PM, Andy Lutomirski wrote:
>
>>> I would suggest that you rework the interface a bit. First a user would
>>>
Hi Andy
On 04/23/2018 02:14 PM, Andy Lutomirski wrote:
> On 04/20/2018 08:55 AM, Eric Dumazet wrote:
>> This patch series provide a new mmap_hook to fs willing to grab
>> a mutex before mm->mmap_sem is taken, to ensure lockdep sanity.
>>
>> This hook allows us to sho
On 04/23/2018 08:58 AM, David Miller wrote:
> From: Yafang Shao
> Date: Sun, 22 Apr 2018 21:50:04 +0800
>
>> With sk_cookie we can identify a socket, that is very helpful for
>> traceing and statistic, i.e. tcp tracepiont and ebpf.
>> So we'd better init it by default for inet socket.
>> When u
On 04/23/2018 01:24 AM, syzbot wrote:
> Hello,
>
> syzbot hit the following crash on net-next commit
> 0638eb573cde5888c0886c7f35da604e5db209a6 (Sat Apr 21 20:06:14 2018 +)
> Merge branch 'ipv6-Another-followup-to-the-fib6_info-change'
> syzbot dashboard link:
> https://syzkaller.appspot.co
ed-off-by: Paul E. McKenney
> > Cc: Ingo Molnar
> Fair enough,
> Acked-by: Peter Zijlstra (Intel)
Yes, I suggested this removal in
https://www.spinics.net/lists/netdev/msg375161.html
Reviewed-by: Eric Dumazet
Thanks Paul.
On 04/21/2018 02:07 AM, Christoph Hellwig wrote:
> On Fri, Apr 20, 2018 at 08:55:38AM -0700, Eric Dumazet wrote:
>> This patch series provide a new mmap_hook to fs willing to grab
>> a mutex before mm->mmap_sem is taken, to ensure lockdep sanity.
>>
>> This hook
s to perform mm operations without delay.
Note that the preparation work (building the array of page
pointers) can also be done from tcp_mmap_hook() while
mmap_sem has not been taken yet, but this is another independent change.
Signed-off-by: Eric Dumazet
---
net/ipv4/tcp.c | 20 +++-
sock_mmap_hook() is the mmap_hook handler provided for socket_file_ops
Following patch will provide tcp_mmap_hook() for TCP protocol.
Signed-off-by: Eric Dumazet
---
include/linux/net.h | 1 +
net/socket.c| 9 +
2 files changed, 10 insertions(+)
diff --git a/include/linux
in multi-threading programs.
Fixes: 93ab6cc69162 ("tcp: implement mmap() for zero copy receive")
Signed-off-by: Eric Dumazet
Reported-by: syzbot
---
include/linux/fs.h | 6 ++
mm/util.c | 19 ++-
2 files changed, 24 insertions(+), 1 deletion(-)
diff --git a/i
This patch series provide a new mmap_hook to fs willing to grab
a mutex before mm->mmap_sem is taken, to ensure lockdep sanity.
This hook allows us to shorten tcp_mmap() execution time (while mmap_sem
is held), and improve multi-threading scalability.
Eric Dumazet (4):
mm: provide a mmap_h
tcp_mmap() execution time
and thus increase mmap() performance in multi-threaded programs.
Fixes: 93ab6cc69162 ("tcp: implement mmap() for zero copy receive")
Signed-off-by: Eric Dumazet
Reported-by: syzbot
---
include/net/tcp.h | 1 +
net/ipv4/af_inet.c | 1 +
net/i
On 04/20/2018 06:57 AM, Jann Horn wrote:
> The old code reads the "opsize" variable from out-of-bounds memory (first
> byte behind the segment) if a broken TCP segment ends directly after an
> opcode that is neither EOL nor NOP.
>
> The result of the read isn't used for anything, so the worst th
On 04/19/2018 09:12 AM, Mikulas Patocka wrote:
>
>
> These bugs are hard to reproduce because vmalloc falls back to kmalloc
> only if memory is fragmented.
>
This sentence is wrong.
because kvmalloc() falls back to vmalloc() ...
On 04/18/2018 02:04 PM, David Ahern wrote:
> On 4/18/18 3:02 PM, syzbot wrote:
>> stack backtrace:
>> CPU: 1 PID: 25 Comm: kworker/1:1 Not tainted 4.16.0+ #5
>> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
>> Google 01/01/2011
>> Workqueue: ipv6_addrconf addrconf_dad_wo
On 04/18/2018 10:55 AM, Michael S. Tsirkin wrote:
> Imagine you want to pass some data to card.
> Natural thing is to just put it in a variable and start DMA.
> However DMA API disallows stack access nowdays,
> so it's natural to put this within struct device.
>
> See e.g.
>
> commit a72
On 04/18/2018 09:44 AM, Mikulas Patocka wrote:
>
>
> On Wed, 18 Apr 2018, Eric Dumazet wrote:
>
>>
>>
>> On 04/18/2018 07:34 AM, Mikulas Patocka wrote:
>>> The patch 74d332c13b21 changes alloc_netdev_mqs to use vzalloc if kzalloc
>>> fails (l
On 04/18/2018 07:34 AM, Mikulas Patocka wrote:
> The patch 74d332c13b21 changes alloc_netdev_mqs to use vzalloc if kzalloc
> fails (later patches change it to kvzalloc).
>
> The problem with this is that if the vzalloc function is actually used,
> virtio_net doesn't work (because it expects tha
On 04/17/2018 09:36 AM, Yafang Shao wrote:
> tcp_rcv_space_adjust is called every time data is copied to user space,
> introducing a tcp tracepoint for which could show us when the packet is
> copied to user.
> This could help us figure out whether there's latency in user process.
>
> When a tcp
On 04/16/2018 08:33 AM, Yafang Shao wrote:
> tcp_rcv_space_adjust is called every time data is copied to user space,
> introducing a tcp tracepoint for which could show us when the packet is
> copied to user.
> This could help us figure out whether there's latency in user process.
>
> When a tcp
On 04/16/2018 02:15 AM, Kirill A. Shutemov wrote:
> On Mon, Apr 16, 2018 at 08:07:09AM +0200, Ingo Molnar wrote:
>>
>> * Eric Dumazet wrote:
>>
>>> Hi Kirill
>>>
>>> For some reason, my hosts instantly crash at boot time, with absolutely no
Hi Kirill
For some reason, my hosts instantly crash at boot time, with absolutely no log
on console.
Bisection pointed to :
$ git bisect bad
194a9749c73d650c0b1dfdee04fb0bdf0a888ba8 is the first bad commit
commit 194a9749c73d650c0b1dfdee04fb0bdf0a888ba8
Author: Kirill A. Shutemov
Date: Mon M
On 04/10/2018 05:53 AM, Matthew Wilcox wrote:
> From: Matthew Wilcox
>
> __GFP_ZERO requests that the object be initialised to all-zeroes,
> while the purpose of a constructor is to initialise an object to a
> particular pattern. We cannot do both. Add a warning to catch any
> users who mista
On 04/09/2018 07:10 AM, Jia-Ju Bai wrote:
> dn_route_init() is never called in atomic context.
>
> The call chain ending up at dn_route_init() is:
> [1] dn_route_init() <- decnet_init()
> decnet_init() is only set as a parameter of module_init().
>
> Despite never getting called from atomic con
On 04/09/2018 07:10 AM, Jia-Ju Bai wrote:
> dccp_init() is never called in atomic context.
> This function is only set as a parameter of module_init().
>
> Despite never getting called from atomic context,
> dccp_init() calls __get_free_pages() with GFP_ATOMIC,
> which waits busily for allocatio
On 04/08/2018 11:06 PM, Dmitry Vyukov wrote:
> On Mon, Apr 9, 2018 at 7:59 AM, syzbot
> wrote:
>> Hello,
>>
>> syzbot hit the following crash on net-next commit
>> 8bde261e535257e81087d39ff808414e2f5aa39d (Sun Apr 1 02:31:43 2018 +)
>> Merge tag 'mlx5-updates-2018-03-30' of
>> git://git.kern
: add ndo_set_rx_mode callback
> implementation for VF")
> Signed-off-by: Gustavo A. R. Silva
> ---
> Changes in v2:
> - Add a null check on a second kmalloc a few lines below. Thanks to
>Eric Dumazet for pointing this out.
>
> drivers/net/ethernet/cavium/thunder/nicvf_main.c
On 04/03/2018 02:29 PM, Gustavo A. R. Silva wrote:
> Add null check on kmalloc() return value in order to prevent
> a null pointer dereference.
>
> Addresses-Coverity-ID: 1467429 ("Dereference null return value")
> Fixes: 37c3347eb247 ("net: thunderx: add ndo_set_rx_mode callback
> implementati
On 04/01/2018 11:31 AM, Anton Gary Ceph wrote:
> As the Linux networking stack is growing, more and more protocols are
> added, increasing the complexity of stack itself.
> Modern processors, contrary to common belief, are very bad in branch
> prediction, so it's our task to give hints to the com
On 03/28/2018 08:51 PM, Dongli Zhang wrote:
> The "BUG_ON(!frag_iter)" in function xenvif_rx_next_chunk() is triggered if
> the received sk_buff is malformed, that is, when the sk_buff has pattern
> (skb->data_len && !skb_shinfo(skb)->nr_frags). Below is a sample call
> stack:
>
>...
>
> The
On 03/28/2018 03:08 AM, Borislav Petkov wrote:
> I guess now that the rdmsr* side does this, you probably should convert
> the wrmsr* side as well.
Yes indeed, thanks for the reminder.
On 03/28/2018 12:30 AM, Saurabh Kr wrote:
> Hi Eric/Angelo,
>
> We are seeing the assertion error in linux kernel 2.4.29 “*kernel: KERNEL:
> assertion (atomic_read(&skb->users) == 0) failed at dev.c(1397)**”.* Based on
> patch provided (_https://patchwork.kernel.org/patch/5368051/_ ) we mer
Commit-ID: 9b9a51354cae933f5640b5bb73bbcd32f989122f
Gitweb: https://git.kernel.org/tip/9b9a51354cae933f5640b5bb73bbcd32f989122f
Author: Eric Dumazet
AuthorDate: Tue, 27 Mar 2018 20:22:33 -0700
Committer: Thomas Gleixner
CommitDate: Wed, 28 Mar 2018 10:34:13 +0200
x86/msr: Make
() to schedule")
Signed-off-by: Eric Dumazet
Reported-by: kbuild test robot
Cc: "H. Peter Anvin"
Cc: Thomas Gleixner
Cc: Ingo Molnar
Cc: Borislav Petkov
---
arch/x86/lib/msr-smp.c | 11 ---
1 file changed, 4 insertions(+), 7 deletions(-)
diff --git a/arch/x86/lib/msr-smp.
microseconds.
> All usage sites are in preemptible context, convert
rdmsr_safe_on_cpu() to
> use a completion instead of busy polling.
> Overall daemon cpu usage was reduced by 35 %, and latencies caused by
> msr_read() disappeared.
> Signed-of
Commit-ID: 67bbd7a8d6bcdc44cc27105ae8c374e9176ceaf1
Gitweb: https://git.kernel.org/tip/67bbd7a8d6bcdc44cc27105ae8c374e9176ceaf1
Author: Eric Dumazet
AuthorDate: Fri, 23 Mar 2018 14:58:18 -0700
Committer: Thomas Gleixner
CommitDate: Tue, 27 Mar 2018 12:01:48 +0200
x86/cpuid: Allow
Commit-ID: 07cde313b2d21f728cec2836db7cdb55476f7a26
Gitweb: https://git.kernel.org/tip/07cde313b2d21f728cec2836db7cdb55476f7a26
Author: Eric Dumazet
AuthorDate: Fri, 23 Mar 2018 14:58:17 -0700
Committer: Thomas Gleixner
CommitDate: Tue, 27 Mar 2018 12:01:47 +0200
x86/msr: Allow
On 03/24/2018 01:09 AM, Ingo Molnar wrote:
>
> * Eric Dumazet wrote:
>
>> I noticed high latencies caused by a daemon periodically reading
>> various MSR on all cpus. KASAN kernels would see ~10ms latencies
>> simply reading one MSR. Even without KASAN, sending
On 03/23/2018 03:17 PM, H. Peter Anvin wrote:
> On 03/23/18 14:58, Eric Dumazet wrote:
>> I noticed high latencies caused by a daemon periodically reading various
>> MSR and cpuid on all cpus. KASAN kernels would see ~10ms latencies
>> simply reading one cpuid. Even without
consume hundreds of usec or more.
Switching to smp_call_function_single_async() and a completion
allows to reschedule and not burn cpu cycles.
Signed-off-by: Eric Dumazet
Cc: "H. Peter Anvin"
Cc: Thomas Gleixner
Cc: Borislav Petkov
Cc: Ingo Molnar
Cc: Hugh Dickins
---
arch/
hundreds of usec.
Converts rdmsr_safe_on_cpu() to use a completion instead
of busy polling.
Overall daemon cpu usage was reduced by 35 %,
and latencies caused by msr_read() disappeared.
Signed-off-by: Eric Dumazet
Cc: "H. Peter Anvin"
Cc: Thomas Gleixner
Cc: Ingo Molnar
Cc: Borislav
On 03/23/2018 02:27 PM, Thomas Gleixner wrote:
>
> Looking at all call sites. None of them is performance critical and all of
> them are in preemptible context.
>
> So we simply can switch the rdmsr_safe_on_cpu() implementation over to wait
> mode completely.
SGTM, thanks for looking, I will se
Fixes: 1da177e4c3f41524 ("Linux-2.6.12-rc2")
Reviewed-by: Eric Dumazet
Thanks Alexander.
consume hundreds of usec or more.
Switching to smp_call_function_single_async() and a completion
allows to reschedule and not burn cpu cycles.
Signed-off-by: Eric Dumazet
Cc: "H. Peter Anvin"
Cc: Thomas Gleixner
Cc: Ingo Molnar
Cc: Hugh Dickins
---
arch/x86/kernel/cp
-by: Eric Dumazet
Cc: "H. Peter Anvin"
Cc: Thomas Gleixner
Cc: Ingo Molnar
Cc: Hugh Dickins
---
v2: fixed the missing part for !CONFIG_SMP
arch/x86/include/asm/msr.h | 6 ++
arch/x86/kernel/msr.c | 2 +-
arch/x86/lib/msr-smp.c | 43 +++
op us a note to
help improve the system]
> url:
https://github.com/0day-ci/linux/commits/Eric-Dumazet/x86-msr-add-rdmsr_safe_on_cpu_resched-and-use-it-in-msr_read/20180319-001007
> config: i386-randconfig-s1-201811 (attached as .config)
> compiler: gcc-6 (Debian 6.4.0-9) 6.
consume hundreds of usec or more.
Switching to smp_call_function_single_async() and a completion
allows to reschedule and not burn cpu cycles.
Signed-off-by: Eric Dumazet
Cc: "H. Peter Anvin"
Cc: Thomas Gleixner
Cc: Ingo Molnar
Cc: Hugh Dickins
---
arch/x86/kernel/cp
-by: Eric Dumazet
Cc: "H. Peter Anvin"
Cc: Thomas Gleixner
Cc: Ingo Molnar
Cc: Hugh Dickins
---
arch/x86/include/asm/msr.h | 1 +
arch/x86/kernel/msr.c | 2 +-
arch/x86/lib/msr-smp.c | 43 ++
3 files changed, 45 insertions(+), 1 deletio
On 03/14/2018 05:16 PM, Eric Dumazet wrote:
>
> typical use after free...
>
> diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
> index
> 190570f21b208d5a17943360a3a6f85e1c2a2187..663e016491773f40f81d9bbfeab3dd68e1c2fc5c
> 100644
> --- a/net/sched/s
On 03/14/2018 04:30 PM, syzbot wrote:
> syzbot has found reproducer for the following crash on net-next commit
> a870a02cc963de35452bbed932560ed69725c4f2 (Tue Mar 13 20:58:39 2018 +)
> pktgen: use dynamic allocation for debug print buffer
>
> So far this crash happened 7 times on mmots, net-
On 03/14/2018 01:11 PM, syzbot wrote:
Hello,
syzbot hit the following crash on net-next commit
be9fc0971a5c27b791608cf9705a04fe96dbd395 (Tue Mar 13 11:44:53 2018 +)
net: fix sysctl_fb_tunnels_only_for_init_net link error
So far this crash happened 2 times on net-next.
Unfortunately, I don
On Wed, Mar 14, 2018 at 7:16 AM, Alexander Potapenko wrote:
>
>
>
> On Wed, Mar 14, 2018 at 3:11 PM Eric Dumazet wrote:
>>
>> On Wed, Mar 14, 2018 at 7:03 AM, Alexander Potapenko
>> wrote:
>> > KMSAN reports use of uninitialized memory in the case when |a
On Wed, Mar 14, 2018 at 7:03 AM, Alexander Potapenko wrote:
> KMSAN reports use of uninitialized memory in the case when |alen| is
> smaller than sizeof(struct netlink_sock), and therefore |nladdr| isn't
> fully copied from the userspace.
>
> Signed-off-by: Alexander Potapenko
> Fixes: 1da177e4c3
ll;
+ else
+ busy_poll = NULL;
do {
rc = 0;
We could instead setup a non NULL netdev_ops pointer on these 'dummy'
devices to not add a check in fast path, but I presume we do
not really care since this fix is for old kernels, and considering how
long it took to discover this bug.
Reviewed-by: Eric Dumazet
On 03/09/2018 03:32 PM, Cong Wang wrote:
On Fri, Mar 9, 2018 at 3:21 PM, Eric Dumazet wrote:
On 03/09/2018 03:05 PM, Cong Wang wrote:
BTW, the warning itself is all about empty names, so perhaps
it's better to fix them separately.
Huh ? You want more syzbot reports ? I do not
501 - 600 of 2745 matches
Mail list logo