[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CH2PR04MB65224683CD7E6C63D60D05BFE7FA0@CH2PR04MB6522.namprd04.prod.outlook.com>
Date: Wed, 25 Nov 2020 11:04:01 +0000
From: Damien Le Moal <Damien.LeMoal@....com>
To: Geert Uytterhoeven <geert@...ux-m68k.org>
CC: Paul Walmsley <paul.walmsley@...ive.com>,
Palmer Dabbelt <palmer@...belt.com>,
Albert Ou <aou@...s.berkeley.edu>,
"linux-riscv@...ts.infradead.org" <linux-riscv@...ts.infradead.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] riscv: defconfig: k210: Disable CONFIG_VT
On 2020/11/25 20:00, Damien Le Moal wrote:
> On 2020/11/25 18:26, Geert Uytterhoeven wrote:
>> Hi Damien,
>>
>> On Wed, Nov 25, 2020 at 10:02 AM Damien Le Moal <Damien.LeMoal@....com> wrote:
>>> On 2020/11/25 17:51, Geert Uytterhoeven wrote:
>>>> On Wed, Nov 25, 2020 at 7:14 AM Damien Le Moal <Damien.LeMoal@....com> wrote:
>>>>> On 2020/11/25 3:57, Geert Uytterhoeven wrote:
>>>>>> There is no need to enable Virtual Terminal support in the Canaan
>>>>>> Kendryte K210 defconfigs, as no terminal devices are supported and
>>>>>> enabled. Hence disable CONFIG_VT, and remove the no longer needed
>>>>>> override for CONFIG_VGA_CONSOLE.
>>>>>>
>>>>>> This reduces kernel size by ca. 65 KiB.
>>>>>
>>>>> Indeed, nice saving. Just tested, and all is good.
>>>>
>>>> I used my old script[1] to check the impact of disabling config options.
>>
>>>> I haven't done enough riscv kernel development yet to assess if I need
>>>> CONFIG_FRAME_POINTER or not.
>>>
>>> Disabling it significantly reduced code size for me. Since the series is more
>>> stable now, it is not really needed, so I disabled it in the defconfig.
>>>
>>> I was just fiddling with CONFIG_UNIX98_PTYS. Disabling it is OK with the simple
>>> busybox userspace (no telnet/xterm like app running). But it saves only about
>>> 1KB with my toolchain (gcc 9.3). So I left that one enabled. I am surprised that
>>> you see 16K size impact... How big is your image ?
>>>
>>> For me, it is 1.768 MB right now for the sdcard defconfig, with CONFIG_VT
>>> disabled and ext2 enabled.
>>
>> It might depend on how you measure. "size" says 15 KiB impact for UNIX98
>> ptys, while bloat-o-meter reported less than 7 (my script uses "size").
>
> I look at the size of the arch/riscv/boot/loader.bin file since that is what
> gets loaded in RAM and booted. It is significantly smaller than vmlinux file
> size. E.g. for the sd card defconfig, I have:
>
> vmlinux: 2369920 B
> loader.bin: 1751250 B
>
>> I'm at 1.88 MiB, with ext4 and without frame pointers.
>> I also got rid of the EFI partition support, and a few I/O schedulers:
>>
>> +CONFIG_PARTITION_ADVANCED=y
>> +# CONFIG_EFI_PARTITION is not set
>> +# CONFIG_MQ_IOSCHED_DEADLINE is not set
>> +# CONFIG_MQ_IOSCHED_KYBER is not set
>
> I have all of these disabled. The schedulers are forced disabled in the sdcard
> defconfig.
>
> I also noticed that it hugely depend on the compiler. Using the buildroot
> generated rv64 gcc 10, the kernel image goes up to almost 2 MB. So for the
> kernel, I keep using the bootlin precompiled gcc 9.3:
>
> https://toolchains.bootlin.com/
>
> Just noticed that they now have a 10.2 version available. Will try it out.
Correction: my PATH was actually pointing to the Fedora riscv64 gcc from the
distro rpm package which is version 10.2, not 9.3.
>
>>
>> Gr{oetje,eeting}s,
>>
>> Geert
>>
>
>
--
Damien Le Moal
Western Digital Research
Powered by blists - more mailing lists