All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jakub Kicinski <kuba@kernel.org>
To: Eric Dumazet <edumazet@google.com>
Cc: Chen Lin <chen45464546@163.com>, Felix Fietkau <nbd@nbd.name>,
	john@phrozen.org, sean.wang@mediatek.com,
	Mark-MC.Lee@mediatek.com, David Miller <davem@davemloft.net>,
	Paolo Abeni <pabeni@redhat.com>,
	Matthias Brugger <matthias.bgg@gmail.com>,
	netdev <netdev@vger.kernel.org>,
	Linux ARM <linux-arm-kernel@lists.infradead.org>,
	linux-mediatek@lists.infradead.org,
	LKML <linux-kernel@vger.kernel.org>,
	Alexander Duyck <alexander.duyck@gmail.com>
Subject: Re: [PATCH v2] net: ethernet: mtk_eth_soc: fix misuse of mem alloc interface netdev[napi]_alloc_frag
Date: Fri, 3 Jun 2022 12:55:16 -0700	[thread overview]
Message-ID: <20220603125516.52353a4a@kernel.org> (raw)
In-Reply-To: <CANn89i+dW+paaybeDkkC0XxYM+Mv_AOnbi6GSLtTgAv9L=TX7Q@mail.gmail.com>

On Fri, 3 Jun 2022 12:11:43 -0700 Eric Dumazet wrote:
> Yes, we only have to review callers and change the documentation and
> implementation.
> 
> The confusion/overhead/generalization came with :
> 
> commit 7ba7aeabbaba484347cc98fbe9045769ca0d118d
> Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Date:   Fri Jun 7 21:20:34 2019 +0200
> 
>     net: Don't disable interrupts in napi_alloc_frag()
> 
>     netdev_alloc_frag() can be used from any context and is used by NAPI
>     and non-NAPI drivers. Non-NAPI drivers use it in interrupt context
>     and NAPI drivers use it during initial allocation (->ndo_open() or
>     ->ndo_change_mtu()). Some NAPI drivers share the same function for the  
>     initial allocation and the allocation in their NAPI callback.
> 
>     The interrupts are disabled in order to ensure locked access from every
>     context to `netdev_alloc_cache'.
> 
>     Let netdev_alloc_frag() check if interrupts are disabled. If they are,
>     use `netdev_alloc_cache' otherwise disable BH and invoke
>     __napi_alloc_frag() for the allocation. The IRQ check is cheaper
>     compared to disabling & enabling interrupts and memory allocation with
>     disabled interrupts does not work on -RT.

Hm, should have looked at the code. Were you thinking of adding a
helper which would replace both netdev_ and napi_ variants and DTRT
internally?

An option for getting GFP_KERNEL in there would be having an rtnl frag
cache. Users who need frags on the reconfig path should be under rtnl,
they can call rtnl_alloc_frag(), which can use GFP_KERNEL internally.
Otherwise the GFP_KERNEL frag cache would need to be protected by
another mutex, I presume. Pre-allocating memory before using the napi
cache seems hard.

_______________________________________________
Linux-mediatek mailing list
Linux-mediatek@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-mediatek

WARNING: multiple messages have this Message-ID (diff)
From: Jakub Kicinski <kuba@kernel.org>
To: Eric Dumazet <edumazet@google.com>
Cc: Chen Lin <chen45464546@163.com>, Felix Fietkau <nbd@nbd.name>,
	john@phrozen.org, sean.wang@mediatek.com,
	Mark-MC.Lee@mediatek.com, David Miller <davem@davemloft.net>,
	Paolo Abeni <pabeni@redhat.com>,
	Matthias Brugger <matthias.bgg@gmail.com>,
	netdev <netdev@vger.kernel.org>,
	Linux ARM <linux-arm-kernel@lists.infradead.org>,
	linux-mediatek@lists.infradead.org,
	LKML <linux-kernel@vger.kernel.org>,
	Alexander Duyck <alexander.duyck@gmail.com>
Subject: Re: [PATCH v2] net: ethernet: mtk_eth_soc: fix misuse of mem alloc interface netdev[napi]_alloc_frag
Date: Fri, 3 Jun 2022 12:55:16 -0700	[thread overview]
Message-ID: <20220603125516.52353a4a@kernel.org> (raw)
In-Reply-To: <CANn89i+dW+paaybeDkkC0XxYM+Mv_AOnbi6GSLtTgAv9L=TX7Q@mail.gmail.com>

On Fri, 3 Jun 2022 12:11:43 -0700 Eric Dumazet wrote:
> Yes, we only have to review callers and change the documentation and
> implementation.
> 
> The confusion/overhead/generalization came with :
> 
> commit 7ba7aeabbaba484347cc98fbe9045769ca0d118d
> Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Date:   Fri Jun 7 21:20:34 2019 +0200
> 
>     net: Don't disable interrupts in napi_alloc_frag()
> 
>     netdev_alloc_frag() can be used from any context and is used by NAPI
>     and non-NAPI drivers. Non-NAPI drivers use it in interrupt context
>     and NAPI drivers use it during initial allocation (->ndo_open() or
>     ->ndo_change_mtu()). Some NAPI drivers share the same function for the  
>     initial allocation and the allocation in their NAPI callback.
> 
>     The interrupts are disabled in order to ensure locked access from every
>     context to `netdev_alloc_cache'.
> 
>     Let netdev_alloc_frag() check if interrupts are disabled. If they are,
>     use `netdev_alloc_cache' otherwise disable BH and invoke
>     __napi_alloc_frag() for the allocation. The IRQ check is cheaper
>     compared to disabling & enabling interrupts and memory allocation with
>     disabled interrupts does not work on -RT.

Hm, should have looked at the code. Were you thinking of adding a
helper which would replace both netdev_ and napi_ variants and DTRT
internally?

An option for getting GFP_KERNEL in there would be having an rtnl frag
cache. Users who need frags on the reconfig path should be under rtnl,
they can call rtnl_alloc_frag(), which can use GFP_KERNEL internally.
Otherwise the GFP_KERNEL frag cache would need to be protected by
another mutex, I presume. Pre-allocating memory before using the napi
cache seems hard.

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

WARNING: multiple messages have this Message-ID (diff)
From: Jakub Kicinski <kuba@kernel.org>
To: Eric Dumazet <edumazet@google.com>
Cc: Chen Lin <chen45464546@163.com>, Felix Fietkau <nbd@nbd.name>,
	john@phrozen.org, sean.wang@mediatek.com,
	Mark-MC.Lee@mediatek.com, David Miller <davem@davemloft.net>,
	Paolo Abeni <pabeni@redhat.com>,
	Matthias Brugger <matthias.bgg@gmail.com>,
	netdev <netdev@vger.kernel.org>,
	Linux ARM <linux-arm-kernel@lists.infradead.org>,
	linux-mediatek@lists.infradead.org,
	LKML <linux-kernel@vger.kernel.org>,
	Alexander Duyck <alexander.duyck@gmail.com>
Subject: Re: [PATCH v2] net: ethernet: mtk_eth_soc: fix misuse of mem alloc interface netdev[napi]_alloc_frag
Date: Fri, 3 Jun 2022 12:55:16 -0700	[thread overview]
Message-ID: <20220603125516.52353a4a@kernel.org> (raw)
In-Reply-To: <CANn89i+dW+paaybeDkkC0XxYM+Mv_AOnbi6GSLtTgAv9L=TX7Q@mail.gmail.com>

On Fri, 3 Jun 2022 12:11:43 -0700 Eric Dumazet wrote:
> Yes, we only have to review callers and change the documentation and
> implementation.
> 
> The confusion/overhead/generalization came with :
> 
> commit 7ba7aeabbaba484347cc98fbe9045769ca0d118d
> Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Date:   Fri Jun 7 21:20:34 2019 +0200
> 
>     net: Don't disable interrupts in napi_alloc_frag()
> 
>     netdev_alloc_frag() can be used from any context and is used by NAPI
>     and non-NAPI drivers. Non-NAPI drivers use it in interrupt context
>     and NAPI drivers use it during initial allocation (->ndo_open() or
>     ->ndo_change_mtu()). Some NAPI drivers share the same function for the  
>     initial allocation and the allocation in their NAPI callback.
> 
>     The interrupts are disabled in order to ensure locked access from every
>     context to `netdev_alloc_cache'.
> 
>     Let netdev_alloc_frag() check if interrupts are disabled. If they are,
>     use `netdev_alloc_cache' otherwise disable BH and invoke
>     __napi_alloc_frag() for the allocation. The IRQ check is cheaper
>     compared to disabling & enabling interrupts and memory allocation with
>     disabled interrupts does not work on -RT.

Hm, should have looked at the code. Were you thinking of adding a
helper which would replace both netdev_ and napi_ variants and DTRT
internally?

An option for getting GFP_KERNEL in there would be having an rtnl frag
cache. Users who need frags on the reconfig path should be under rtnl,
they can call rtnl_alloc_frag(), which can use GFP_KERNEL internally.
Otherwise the GFP_KERNEL frag cache would need to be protected by
another mutex, I presume. Pre-allocating memory before using the napi
cache seems hard.

  reply	other threads:[~2022-06-03 19:55 UTC|newest]

Thread overview: 54+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-06-03  4:10 [PATCH] net: ethernet: mtk_eth_soc: fix misuse of mem alloc interface netdev_alloc_frag Chen Lin
2022-06-03  4:10 ` Chen Lin
2022-06-03  4:10 ` Chen Lin
2022-06-03  4:24 ` Jakub Kicinski
2022-06-03  4:24   ` Jakub Kicinski
2022-06-03  4:24   ` Jakub Kicinski
2022-06-03  4:30 ` Felix Fietkau
2022-06-03  4:30   ` Felix Fietkau
2022-06-03  4:30   ` Felix Fietkau
2022-06-03  8:46   ` [PATCH v2] net: ethernet: mtk_eth_soc: fix misuse of mem alloc interface netdev[napi]_alloc_frag Chen Lin
2022-06-03  8:46     ` Chen Lin
2022-06-03  8:46     ` Chen Lin
2022-06-03 15:25     ` Alexander Duyck
2022-06-03 15:25       ` Alexander Duyck
2022-06-03 15:25       ` Alexander Duyck
2022-06-03 15:33       ` Alexander Duyck
2022-06-03 15:33         ` Alexander Duyck
2022-06-03 15:33         ` Alexander Duyck
2022-06-05  2:22         ` Chen Lin
2022-06-05  2:22           ` Chen Lin
2022-06-05  2:22           ` Chen Lin
2022-06-05  3:12         ` [PATCH v3] " Chen Lin
2022-06-05  3:12           ` Chen Lin
2022-06-05  3:12           ` Chen Lin
2022-06-06 21:34           ` Jakub Kicinski
2022-06-06 21:34             ` Jakub Kicinski
2022-06-06 21:34             ` Jakub Kicinski
2022-06-06 23:39             ` [PATCH v4] " Chen Lin
2022-06-06 23:39               ` Chen Lin
2022-06-06 23:39               ` Chen Lin
2022-06-07 23:14               ` Jakub Kicinski
2022-06-07 23:14                 ` Jakub Kicinski
2022-06-07 23:14                 ` Jakub Kicinski
2022-06-08 12:43                 ` Chen Lin
2022-06-08 12:43                   ` Chen Lin
2022-06-08 12:43                   ` Chen Lin
2022-06-08 12:46                 ` [PATCH v5] " Chen Lin
2022-06-08 12:46                   ` Chen Lin
2022-06-08 12:46                   ` Chen Lin
2022-06-09  3:50                   ` patchwork-bot+netdevbpf
2022-06-09  3:50                     ` patchwork-bot+netdevbpf
2022-06-09  3:50                     ` patchwork-bot+netdevbpf
2022-06-03 17:25     ` [PATCH v2] " Eric Dumazet
2022-06-03 17:25       ` Eric Dumazet
2022-06-03 17:25       ` Eric Dumazet
2022-06-03 18:59       ` Jakub Kicinski
2022-06-03 18:59         ` Jakub Kicinski
2022-06-03 18:59         ` Jakub Kicinski
2022-06-03 19:11         ` Eric Dumazet
2022-06-03 19:11           ` Eric Dumazet
2022-06-03 19:11           ` Eric Dumazet
2022-06-03 19:55           ` Jakub Kicinski [this message]
2022-06-03 19:55             ` Jakub Kicinski
2022-06-03 19:55             ` Jakub Kicinski

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20220603125516.52353a4a@kernel.org \
    --to=kuba@kernel.org \
    --cc=Mark-MC.Lee@mediatek.com \
    --cc=alexander.duyck@gmail.com \
    --cc=chen45464546@163.com \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=john@phrozen.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mediatek@lists.infradead.org \
    --cc=matthias.bgg@gmail.com \
    --cc=nbd@nbd.name \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    --cc=sean.wang@mediatek.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.