linux-kernel - Re: [RESEND PATCH 5.10.x] mtd: spinand: Fix incorrect parameters for on-die ECC

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

Date:   Mon, 30 Aug 2021 11:18:50 +0200
From:   Frieder Schrempf <frieder.schrempf@...tron.de>
To:     Miquel Raynal <miquel.raynal@...tlin.com>,
        Frieder Schrempf <frieder@...s.de>
Cc:     stable@...r.kernel.org,
        voice INTER connect GmbH <developer@...ceinterconnect.de>,
        Alexander Lobakin <alobakin@...me>,
        Felix Fietkau <nbd@....name>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        linux-kernel@...r.kernel.org, linux-mtd@...ts.infradead.org,
        Richard Weinberger <richard@....at>,
        YouChing Lin <ycllin@...c.com.tw>
Subject: Re: [RESEND PATCH 5.10.x] mtd: spinand: Fix incorrect parameters for
 on-die ECC

On 30.08.21 10:41, Miquel Raynal wrote:
> Hi Frieder,
> 
> Frieder Schrempf <frieder@...s.de> wrote on Mon, 30 Aug 2021 09:21:07
> +0200:
> 
>> From: Frieder Schrempf <frieder.schrempf@...tron.de>
>>
>> The new generic NAND ECC framework stores the configuration and requirements
>> in separate places since commit 93ef92f6f422 (" mtd: nand: Use the new generic ECC object ").
>> In 5.10.x The SPI NAND layer still uses only the requirements to track the ECC
>> properties. This mismatch leads to values of zero being used for ECC strength
>> and step_size in the SPI NAND layer wherever nanddev_get_ecc_conf() is used and
>> therefore breaks the SPI NAND on-die ECC support in 5.10.x.
>>
>> By using nanddev_get_ecc_requirements() instead of nanddev_get_ecc_conf() for
>> SPI NAND, we make sure that the correct parameters for the detected chip are
>> used. In later versions (5.11.x) this is fixed anyway with the implementation
>> of the SPI NAND on-die ECC engine.
>>
>> Cc: stable@...r.kernel.org # 5.10.x
>> Reported-by: voice INTER connect GmbH <developer@...ceinterconnect.de>
>> Signed-off-by: Frieder Schrempf <frieder.schrempf@...tron.de>
> 
> Why not just reverting 9a333a72c1d0 ("mtd: spinand: Use
> nanddev_get_ecc_conf() when relevant")? I think using this "new"
> nanddev_get_ecc_requirements() helper because it fits the purpose even
> if it is wrong [1] doesn't bring the right information. I know it is
> strictly equivalent but I would personally prefer keeping the old
> writing "nand->eccreq.xxxx".

I think reverting 9a333a72c1d0 to use nand->eccreq.xxxx doesn't work as the eccreq member has already been removed in 93ef92f6f422 ("mtd: nand: Use the new generic ECC object"). So we would need to revert this commit, too. That would work for the SPI NAND layer, but might have implications on RAW NAND as it already uses the new generic ECC object with 'ctx.conf' and 'requirements'. At least I can't tell how this would affect the RAW NAND layer.

> 
> [1] We don't want the requirements but the actual current configuration
> here, which was the primary purpose of the initial patch which ended
> being a mistake at that point in time because the SPI-NAND core was not
> ready yet.
> 
> Thanks,
> Miquèl