linux-kernel - Re: [PATCH v5] erofs: use Z_EROFS_LCLUSTER_TYPE

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]

Message-ID: <18767765-53b5-4e78-b50d-9305fe1cb2d0@linux.alibaba.com>
Date: Mon, 17 Mar 2025 14:15:16 +0800
From: Gao Xiang <hsiangkao@...ux.alibaba.com>
To: Chao Yu <chao@...nel.org>
Cc: linux-kernel@...r.kernel.org, Hongzhen Luo <hongzhen@...ux.alibaba.com>,
 linux-erofs mailing list <linux-erofs@...ts.ozlabs.org>
Subject: Re: [PATCH v5] erofs: use Z_EROFS_LCLUSTER_TYPE_MAX to simplify
 switches

Hi Chao,

On 2025/3/17 14:03, Chao Yu wrote:
> On 3/17/25 01:17, Gao Xiang wrote:
>> Hi Chao,
>>

...

>>
>> Previously, it was useful before Z_EROFS_LCLUSTER_TYPE_HEAD2 was
>> introduced, but the `default:` case is already deadcode now.
> 
> Xiang, thanks for the explanation.
> 
> So seems it can happen when mounting last image w/ old kernel which can not
> support newly introduced Z_EROFS_LCLUSTER_TYPE_* type, then it makes sense to
> return EOPNOTSUPP.

Yeah.

> 
>>
>>>
>>> Btw, we'd better to do sanity check for m->type in z_erofs_load_full_lcluster(),
>>> then we can treat m->type as reliable variable later.
>>>
>>>       advise = le16_to_cpu(di->di_advise);
>>>       m->type = advise & Z_EROFS_LI_LCLUSTER_TYPE_MASK;
>>>       if (m->type >= Z_EROFS_LCLUSTER_TYPE_MAX) {
>>
>> It's always false here.
> 
> So, what do you think of this?
> 
>  From af584b2eacd468f145e9ee31ccdeedb7355d5afd Mon Sep 17 00:00:00 2001
> From: Chao Yu <chao@...nel.org>
> Date: Mon, 17 Mar 2025 13:57:55 +0800
> Subject: [PATCH] erofs: remove dead codes for cleanup
> 
> z_erofs_extent_lookback() and z_erofs_get_extent_decompressedlen() tries
> to do sanity check on m->type, however their caller z_erofs_map_blocks_fo()
> has already checked that, so let's remove those dead codes.

z_erofs_extent_lookback() will (lookback) read new lcn in
z_erofs_load_lcluster_from_disk() so it won't be covered by
the original z_erofs_map_blocks_fo().

I think this check can be resolved in
z_erofs_load_lcluster_from_disk() instead but maybe address
for the next cycle? since there are already enough features
for this cycle and I have to make sure no major issues....

Thanks,
Gao Xiang