linux-kernel - Re: [PATCH] scsi: scsi_error: the Error Handler base on SCSI Device

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: <b62930b3-9b80-4b06-b922-c38c7e309048@acm.org>
Date: Tue, 25 Nov 2025 08:36:22 -0800
From: Bart Van Assche <bvanassche@....org>
To: JiangJianJun <jiangjianjun3@...wei.com>,
 James.Bottomley@...senPartnership.com, martin.petersen@...cle.com,
 linux-scsi@...r.kernel.org, linux-kernel@...r.kernel.org
Cc: hare@...e.de, dlemoal@...nel.org, yangxingui@...wei.com,
 hewenliang4@...wei.com, yangyun50@...wei.com, wuyifeng10@...wei.com,
 wubo40@...wei.com
Subject: Re: [PATCH] scsi: scsi_error: the Error Handler base on SCSI Device


On 11/25/25 5:48 AM, JiangJianJun wrote:
> When a SCSI device fails, SCSI locks the host and enters an error handler,
> which causes all I/O operations on the host to be locked. This performance
> impact is even more pronounced when a large number of devices are connected
> to the same host. So I believe it's necessary to downgrade the large lock.
> 
> This commit binds an Error Handler to the device, so that when a device
> failure occurs, only the current device is locked. If the device fails to
> recover, the default Error Handler will still be activated.

The current behavior should be retained as the default and the new
behavior should only be enabled if a SCSI LLD requests the new behavior
because many SCSI LLD .eh_host_reset_handler implementations are based
on the assumption that no I/O is in progress if they are called.

Bart.