linux-kernel - Re: [PATCH v2 0/5] mtd: spi-nor: Add support for Octal 8D-8D-8D mode

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20200515065508.6cra7nwt3jpxwvr2@yadavpratyush.com>
Date:   Fri, 15 May 2020 12:25:08 +0530
From:   Pratyush Yadav <me@...avpratyush.com>
To:     masonccyang@...c.com.tw
Cc:     vigneshr@...com, tudor.ambarus@...rochip.com, juliensu@...c.com.tw,
        richard@....at, miquel.raynal@...tlin.com,
        linux-kernel@...r.kernel.org, linux-spi@...r.kernel.org,
        broonie@...nel.org, linux-mtd@...ts.infradead.org,
        Boris Brezillon <boris.brezillon@...labora.com>,
        Pratyush Yadav <p.yadav@...com>
Subject: Re: [PATCH v2 0/5] mtd: spi-nor: Add support for Octal 8D-8D-8D mode

Hi Mason,

On 15/05/20 10:26AM, masonccyang@...c.com.tw wrote:
> 
> Hi Pratyush,
> 
> > > > > I can't apply your patches to enable xSPI Octal mode for 
> > > > > mx25uw51245g because your patches set up Octal protocol first and 
> > > > > then using Octal protocol to write Configuration Register 2(CFG 
> > > > > Reg2). I think driver
> > > > > should write CFG Reg2 in SPI 1-1-1 mode (power on state) and make 
> sure
> > > > > write CFG Reg 2 is success and then setup Octa protocol in the 
> last.
> > > > 
> > > > Register writes should work in 1S mode, because nor->reg_proto is 
> only 
> > > > set _after_ 8D mode is enabled (see spi_nor_octal_dtr_enable()). In 
> > > > fact, both patch 15 and 16 in my series use register writes in 1S 
> mode.
> > > 
> > > but I didn't see driver roll back "nor->read/write_proto = 1" 
> > > if xxx->octal_dtr_enable() return failed!
> > 
> > I copied what spi_nor_quad_enable() did, and made failure fatal. So if 
> > xxx->octal_dtr_enable() fails, the probe would fail and the flash would 
> > be unusable. You can try your hand at a fallback system where you try 
> 
> IMHO, it's not a good for system booting from SPI-NOR, 
> driver should still keep system alive in SPI 1-1-1 mode in case of 
> enable Octal/Quad failed.

I agree.
 
> Therefore, my patches is to setup nor->read/write_proto = 8 in case 
> driver enable Octal mode is success. And to enable Octal mode in
> spi_nor_late_init_params()rather than as spi_nor_quad_enable()did.

Like I mentioned before, spi_nor_late_init_params() is called _before_ 
we call spi_nor_spimem_adjust_hwcaps(). That call is needed to make sure 
the controller also supports octal mode operations. Otherwise, you'd end 
up enabling octal mode on a controller that doesn't support it with no 
way of going back now.

But we can still have a fallback mechanism even in spi_nor_init() that 
would switch to a "less preferred" mode (like 1-1-1 mode) if "more 
preferred" ones like octal or quad fail.

That said, I think it would be a good idea to not keep tacking features 
on this series. This makes it harder for reviewers because now they are 
trying to shoot a moving target. Let basic 8D support stabilize and get 
merged in, and then a fallback system can be submitted as a separate 
patch series.

-- 
Regards,
Pratyush Yadav