Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   "Fabio M. De Francesco" <fmdefrancesco@gmail.com>
To:     Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>,
        Ira Weiny <ira.weiny@intel.com>
Cc:     Alexander Duyck <alexander.duyck@gmail.com>,
        Jakub Kicinski <kuba@kernel.org>,
        "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Netdev <netdev@vger.kernel.org>,
        Maciej Fijalkowski <maciej.fijalkowski@intel.com>,
        Jesper Dangaard Brouer <hawk@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        intel-wired-lan <intel-wired-lan@lists.osuosl.org>,
        Alexander Duyck <alexanderduyck@fb.com>,
        John Fastabend <john.fastabend@gmail.com>,
        Jesse Brandeburg <jesse.brandeburg@intel.com>,
        Alexei Starovoitov <ast@kernel.org>, bpf <bpf@vger.kernel.org>,
        Paolo Abeni <pabeni@redhat.com>,
        LKML <linux-kernel@vger.kernel.org>,
        Tony Nguyen <anthony.l.nguyen@intel.com>
Subject: Re: [Intel-wired-lan] [PATCH] ixgbe: Use kmap_local_page in ixgbe_check_lbtest_frame()
Date:   Sat, 01 Oct 2022 00:03:24 +0200
Message-ID: <832292081.0ifERbkFSE@localhost.localdomain>
In-Reply-To: <27280395.gRfpFWEtPU@localhost.localdomain>
References: <20220629085836.18042-1-fmdefrancesco@gmail.com> <22aa8568-7f6e-605e-7219-325795b218b7@intel.com> <27280395.gRfpFWEtPU@localhost.localdomain>
MIME-Version: 1.0
Content-Transfer-Encoding: 7Bit
Content-Type: text/plain; charset="us-ascii"
Precedence: bulk

On Friday, September 23, 2022 5:05:43 PM CEST Fabio M. De Francesco wrote:
> Hi Anirudh,
> 
> On Friday, September 23, 2022 12:38:02 AM CEST Anirudh Venkataramanan wrote:
> > On 9/22/2022 1:58 PM, Alexander Duyck wrote:
> > > On Thu, Sep 22, 2022 at 1:07 PM Anirudh Venkataramanan
> > > <anirudh.venkataramanan@intel.com> wrote:

[snip]

> > Is using page_address() directly beneficial in some way?
> 
> A possible call chain on 32 bits kernels is the following:
> 
> kmap_local_page() ->
>  __kmap_local_page_prot() { 
> 	if (!IS_ENABLED(CONFIG_DEBUG_KMAP_LOCAL_FORCE_MAP) && |
> PageHighMem(page))
> 		return page_address(page);
> 
> ....
> }
> 
> How many instructions can you save calling page_address() directly?
> If you don't know, look at the assembly.

I just realized that perhaps you were expecting something like either "No, it 
is not directly beneficial because []" or "Yes, it is directly beneficial 
because []".

Instead, I used a rhetoric question that might not have been so clear as I 
thought. This kind of construct is so largely used in my native language, that 
nobody might misunderstand. I'm not so sure if it is the same in English.

I mean, are those dozen "unnecessary" further assembly instructions too many 
or too few to care about? I _think_ that they are too many.

Therefore, by showing a possible call chain in 32 bits architectures, I 
indirectly responded "no, I can't see any direct benefit", at least because....

1) Whatever the architecture, if pages can't come from Highmem, code always 
ends up calling page_address(). In 32 bits archs they waste precious kernel 
stack space (a scarce resources) only to build two stack frames (one per each 
called functions).

 2) Developers adds further work to the CPU and force the kernel to run 
unnecessary code.

I'll always use page_address() when I can "prove" that the allocation cannot 
come from ZONE_HIGHMEM.

Thanks,

Fabio