Received: from eggs.gnu.org ([209.51.188.92]:43595) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1h1puQ-00036L-Tt for qemu-devel@nongnu.org; Thu, 07 Mar 2019 05:05:53 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1h1puP-0006rW-Us for qemu-devel@nongnu.org; Thu, 07 Mar 2019 05:05:50 -0500 From: Markus Armbruster References: <20190307093723.655-1-armbru@redhat.com> <20190307093723.655-3-armbru@redhat.com> Date: Thu, 07 Mar 2019 11:05:41 +0100 In-Reply-To: <20190307093723.655-3-armbru@redhat.com> (Markus Armbruster's message of "Thu, 7 Mar 2019 10:37:21 +0100") Message-ID: <87h8cft2x6.fsf@dusky.pond.sub.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [RFC PATCH v6 2/4] hw/block: Pad undersized read-only images with 0xFF List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: alex.bennee@linaro.org, lersek@redhat.com, philmd@redhat.com, kwolf@redhat.com, mreitz@redhat.com, qemu-block@nongnu.org Markus Armbruster writes: > From: Alex Benn=C3=A9e > > We reject undersized images. As of the previous commit, even with a > decent error message. Still, this is a potentially confusing > stumbling block when you move from using -bios to using -drive > if=3Dpflash,file=3Dblob,format=3Draw,readonly for loading your firmware > code. To mitigate that we automatically pad in the read-only case and > warn the user when we have performed magic to enable things to Just > Work (tm). > > Signed-off-by: Alex Benn=C3=A9e > Reviewed-by: Laszlo Ersek > Signed-off-by: Markus Armbruster I think this convenience feature is a bad idea, and this patch should not be applied. Here are my reasons. 1. As I explained in my disccussion of v5[*], there is no single true way to pad. This patch pads with 0xFF. When working with physical devices, you'd sometimes pad that way, but at other times, you'd pad differently. 2. Except this patch doesn't *actually* pad with 0xFF. The block layer silently pads with zeros up to the next multiple of 512. Evidence: $ yes | dd of=3D4090b.img bs=3D1 count=3D4090 4090+0 records in 4090+0 records out 4090 bytes (4.1 kB, 4.0 KiB) copied, 0.0186459 s, 219 kB/s $ qemu-io -f raw -c 'read -v 4000 96' 4090b.img 00000fa0: 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a y.y.y.y.y.y= .y.y. 00000fb0: 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a y.y.y.y.y.y= .y.y. 00000fc0: 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a y.y.y.y.y.y= .y.y. 00000fd0: 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a y.y.y.y.y.y= .y.y. 00000fe0: 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a 79 0a y.y.y.y.y.y= .y.y. 00000ff0: 79 0a 79 0a 79 0a 79 0a 79 0a 00 00 00 00 00 00 y.y.y.y.y..= ..... read 96/96 bytes at offset 4000 96 bytes, 1 ops; 0.0001 sec (694.444 KiB/sec and 7407.4074 ops/sec) This patch then pads some more with 0xFF to the flash memory size. Because of that the way the magic padding works makes no sense, to be frank. Going back to v3's zero padding would at least be explainable without blushing. I consider the block layer's padding a misfeature here, but right now we got to play the hand we've been dealt. 3. Convenience magic has bitten us in the posterior so many times. Just say no unless there's a really compelling use case. Where's the use case for this one? We've rejected undersized images for ages, and nobody complained. Why add convenience magic now? [*] Message-ID: <87r2bl5oah.fsf@dusky.pond.sub.org> https://lists.nongnu.org/archive/html/qemu-devel/2019-03/msg01115.html