>-Original Message-
>From: Eric Dumazet
>Sent: Tuesday, November 3, 2020 8:16 PM
>To: Vladimir Oltean ; Claudiu Manoil
>
>Cc: Jakub Kicinski ; netdev@vger.kernel.org; David S .
>Miller ; james.jur...@ametek.com
>Subject: Re: [PATCH net v2 1/2] gianfar: Replace s
On 11/3/20 6:49 PM, Vladimir Oltean wrote:
> On Tue, Nov 03, 2020 at 05:41:36PM +, Claudiu Manoil wrote:
>> This is the patch:
>> https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=d145c9031325fed963a887851d9fa42516efd52b
>>
>> are you sure you have it applied?
>
> Ac
On Tue, Nov 03, 2020 at 07:49:06PM +0200, Vladimir Oltean wrote:
> On Tue, Nov 03, 2020 at 05:41:36PM +, Claudiu Manoil wrote:
> > This is the patch:
> > https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=d145c9031325fed963a887851d9fa42516efd52b
> >
> > are you sure you
On Tue, Nov 03, 2020 at 09:36:55AM -0800, Jakub Kicinski wrote:
> IIRC we did this because too many drivers used dev_kfree_skb
> incorrectly and made the dropwatch output very noisy.
Nice, so that's why the drop monitor never complains with my misplaced
dev_kfree_skb_any calls...
On Tue, Nov 03, 2020 at 05:41:36PM +, Claudiu Manoil wrote:
> This is the patch:
> https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=d145c9031325fed963a887851d9fa42516efd52b
>
> are you sure you have it applied?
Actually? No, I didn't have it applied... I had thought t
>-Original Message-
>From: Vladimir Oltean
>Sent: Tuesday, November 3, 2020 7:30 PM
>To: Claudiu Manoil
>Cc: Jakub Kicinski ; netdev@vger.kernel.org; David S .
>Miller ; james.jur...@ametek.com
>Subject: Re: [PATCH net v2 1/2] gianfar: Replace skb_realloc_headroom w
On Tue, 3 Nov 2020 09:36:55 -0800 Jakub Kicinski wrote:
> On Tue, 3 Nov 2020 19:30:07 +0200 Vladimir Oltean wrote:
> > On Tue, Nov 03, 2020 at 05:18:25PM +, Claudiu Manoil wrote:
> > > It's either the dev_kfree_skb_any from the dma mapping error path or the
> > > one
> > > from skb_cow_head(
On Tue, Nov 03, 2020 at 07:24:41PM +0200, Vladimir Oltean wrote:
> On Tue, Nov 03, 2020 at 09:12:26AM -0800, Jakub Kicinski wrote:
> > While it is something to be fixed - is anything other than pktgen
> > making use of IFF_TX_SKB_SHARING? Are you running pktgen, Vladimir?
>
> Nope, just iperf3 TCP
On Tue, 3 Nov 2020 19:30:07 +0200 Vladimir Oltean wrote:
> On Tue, Nov 03, 2020 at 05:18:25PM +, Claudiu Manoil wrote:
> > It's either the dev_kfree_skb_any from the dma mapping error path or the one
> > from skb_cow_head()'s error path. A confirmation would help indeed.
>
> It says "consum
On Tue, Nov 03, 2020 at 05:18:25PM +, Claudiu Manoil wrote:
> It's either the dev_kfree_skb_any from the dma mapping error path or the one
> from skb_cow_head()'s error path. A confirmation would help indeed.
It says "consume", not "kfree", which in my mind would make it point
towards the onl
On Tue, Nov 03, 2020 at 07:18:49PM +0200, Vladimir Oltean wrote:
> On Tue, Nov 03, 2020 at 06:36:30PM +0200, Julian Wiedmann wrote:
> > Given the various skb modifications in its xmit path, I wonder why
> > gianfar doesn't clear IFF_TX_SKB_SHARING.
>
> Thanks for the hint Julian :) I'll try to see
On Tue, Nov 03, 2020 at 09:12:26AM -0800, Jakub Kicinski wrote:
> While it is something to be fixed - is anything other than pktgen
> making use of IFF_TX_SKB_SHARING? Are you running pktgen, Vladimir?
Nope, just iperf3 TCP and PTP. The problem is actually with PTP, I've
been testing gianfar with
On Tue, Nov 03, 2020 at 06:36:30PM +0200, Julian Wiedmann wrote:
> Given the various skb modifications in its xmit path, I wonder why
> gianfar doesn't clear IFF_TX_SKB_SHARING.
Thanks for the hint Julian :) I'll try to see if it makes any
difference.
Just a wild guess, but maybe because the gian
>-Original Message-
>From: Jakub Kicinski
>Sent: Tuesday, November 3, 2020 6:31 PM
>To: Vladimir Oltean
>Cc: Claudiu Manoil ; netdev@vger.kernel.org;
>David S . Miller ; james.jur...@ametek.com
>Subject: Re: [PATCH net v2 1/2] gianfar: Replace skb_realloc_headroom w
On Tue, 3 Nov 2020 17:08:43 + Claudiu Manoil wrote:
> >> Still crashes for me:
> >>
> >
> >Given the various skb modifications in its xmit path, I wonder why
> >gianfar doesn't clear IFF_TX_SKB_SHARING.
>
> Hi Vladimir,
> Can you try the above suggestion on your setup?
While it is somethi
>-Original Message-
>From: Julian Wiedmann
>Sent: Tuesday, November 3, 2020 6:37 PM
>To: Vladimir Oltean ; Claudiu Manoil
>
>Cc: netdev@vger.kernel.org; David S . Miller ; Jakub
>Kicinski ; james.jur...@ametek.com
>Subject: Re: [PATCH net v2 1/2] gianfar: Replace s
his means
>> that the skb provided by the stack is required to have
>> enough headroom otherwise a new skb needs to be created
>> by the driver to accommodate the timestamp inserted by h/w.
>> Up until now the driver was relying on skb_realloc_headroom()
>> to create
On Tue, 3 Nov 2020 18:13:19 +0200 Vladimir Oltean wrote:
> [14538.046926] PC is at skb_release_data+0x6c/0x14c
> [14538.051518] LR is at consume_skb+0x38/0xd8
> [14538.055588] pc : []lr : []psr: 200f0013
> [14538.061817] sp : c28f1da8 ip : fp : c265aa40
> [14538.067010] r10:
ired to have
> enough headroom otherwise a new skb needs to be created
> by the driver to accommodate the timestamp inserted by h/w.
> Up until now the driver was relying on skb_realloc_headroom()
> to create new skbs to accommodate PTP frames. Turns out that
> this method is not reliab
ave
> enough headroom otherwise a new skb needs to be created
> by the driver to accommodate the timestamp inserted by h/w.
> Up until now the driver was relying on skb_realloc_headroom()
> to create new skbs to accommodate PTP frames. Turns out that
> this method is not reliable i
accommodate the timestamp inserted by h/w.
Up until now the driver was relying on skb_realloc_headroom()
to create new skbs to accommodate PTP frames. Turns out that
this method is not reliable in this context at least, as
skb_realloc_headroom() for PTP frames can cause random crashes,
mostly in
From: Herbert Xu <[EMAIL PROTECTED]>
Date: Sun, 14 Oct 2007 22:49:04 +0800
> [IPV6]: Avoid skb_copy/pskb_copy/skb_realloc_headroom on input
>
> This patch replaces unnecessary uses of skb_copy by pskb_expand_head
> on the IPv6 input path.
>
> This allows us to remove t
[IPV6]: Avoid skb_copy/pskb_copy/skb_realloc_headroom on input
This patch replaces unnecessary uses of skb_copy by pskb_expand_head
on the IPv6 input path.
This allows us to remove the double pointers later.
Signed-off-by: Herbert Xu <[EMAIL PROTECTED]>
---
net/ipv6/exthdrs.c
[NETFILTER]: Avoid skb_copy/pskb_copy/skb_realloc_headroom
This patch replaces unnecessary uses of skb_copy, pskb_copy and
skb_realloc_headroom by functions such as skb_make_writable and
pskb_expand_head.
This allows us to remove the double pointers later.
Signed-off-by: Herbert Xu <[EM
radio_hard_start_xmit will do (simplified):
1) ajust the headroom space
hh_len = LL_RESERVED_SPACE(bdev);
if (unlikely(skb_headroom(skb) < hh_len && bdev->hard_header)) {
struct sk_buff *skb2;
skb2 = skb_realloc_headroom(skb, LL_RESERVED_SPACE(dev));
if
25 matches
Mail list logo
