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>] [day] [month] [year] [list]
Date:   Wed, 17 Aug 2022 23:20:08 +0800
From:   John Garry <john.garry@...wei.com>
To:     <damien.lemoal@...nsource.wdc.com>, <oliver.sang@...el.com>
CC:     <linux-ide@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <linux-scsi@...r.kernel.org>, <martin.petersen@...cle.com>,
        <hch@....de>, John Garry <john.garry@...wei.com>
Subject: [RFT PATCH] ata: libata: Set __ATA_BASE_SHT max_sectors

Commit 0568e6122574 ("ata: libata-scsi: cap ata_device->max_sectors
according to shost->max_sectors") inadvertently capped the max_sectors
value for some SATA disks to a value which is lower than we would want.

For a device which supports LBA48, we would previously have request queue
max_sectors_kb and max_hw_sectors_kb values of 1280 and 32767 respectively.

For AHCI controllers, the value chosen for shost max sectors comes from
the minimum of the SCSI host default max sectors in
SCSI_DEFAULT_MAX_SECTORS (1024) and the shost DMA device mapping limit.

This means that we would now set the max_sectors_kb and max_hw_sectors_kb
values for a disk which supports LBA48 at 512, ignoring DMA mapping limit.

As report by Oliver at [0], this caused a performance regression.

Fix by picking a large enough max sectors value for ATA host controllers
such that we don't needlessly reduce max_sectors_kb for LBA48 disks.

[0] https://lore.kernel.org/linux-ide/YvsGbidf3na5FpGb@xsang-OptiPlex-9020/T/#m22d9fc5ad15af66066dd9fecf3d50f1b1ef11da3

Fixes: 0568e6122574 ("ata: libata-scsi: cap ata_device->max_sectors according to shost->max_sectors")
Reported-by: Oliver Sang <oliver.sang@...el.com>
Signed-off-by: John Garry <john.garry@...wei.com>

diff --git a/include/linux/libata.h b/include/linux/libata.h
index 0269ff114f5a..698032e5ef2d 100644
--- a/include/linux/libata.h
+++ b/include/linux/libata.h
@@ -1382,7 +1382,8 @@ extern const struct attribute_group *ata_common_sdev_groups[];
 	.proc_name		= drv_name,			\
 	.slave_destroy		= ata_scsi_slave_destroy,	\
 	.bios_param		= ata_std_bios_param,		\
-	.unlock_native_capacity	= ata_scsi_unlock_native_capacity
+	.unlock_native_capacity	= ata_scsi_unlock_native_capacity,\
+	.max_sectors		= ATA_MAX_SECTORS_LBA48
 
 #define ATA_SUBBASE_SHT(drv_name)				\
 	__ATA_BASE_SHT(drv_name),				\
-- 
2.35.3

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ