Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 29 Dec 2020 13:08:14 +0000
From:      bugzilla-noreply@freebsd.org
To:        bugs@FreeBSD.org
Subject:   [Bug 224496] mpr and mps drivers seems to have issues with large seagate drives
Message-ID:  <bug-224496-227-UqiOaxnwlr@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-224496-227@https.bugs.freebsd.org/bugzilla/>
References:  <bug-224496-227@https.bugs.freebsd.org/bugzilla/>

next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D224496

--- Comment #47 from elij <elij+freebsd@wundrd.net> ---
I have a SAS9305-16i, and was seeing similar errors to some of the folks he=
re.
The comment about NCQ got me thinking, and sure enough.. looks like broadcom
had a new firmware out (I had 16.00.11.00 where I saw the errors).

16.00.12.00 has these two as the fixed "defect list":

    ID:
        DCSG00398894
    Headline: SATA only:
        WRITE SAME NCQ encapsulation assumes NonData NCQ is supported if Ze=
ro
        EXT is Supported
    Description Of Change:
        Disable NCQ encapsulation if Zero EXT is supported but Non Data NCQ=
 is
        not supported
    Issue Description:
        In case if Zero EXT is supported but Non Data NCQ is not supported =
by
        drive, WRITE SAME NCQ encapsulation would send Non Data NCQ command=
 to
        the drive.
        Drive would fail the command as Non Data NCQ is not supported by dr=
ive.
        This will cause command failure to host.
    Steps To Reproduce:
        IO errors are observed when mkfs.ext4 operation is done on drives t=
hat
        support Zero EXT but do not support Non Data NCQ.


    ID:=20
        DCSG00411882 (Port Of Defect DCSG00139294)
    Headline:
        SATA only : Controller hang is observed due to recursive function c=
all.
    Description Of Change:
        Avoid starting of pended commands to SATA drive from completion
        functions of a command to avoid recursion.
    Issue Description:
        Controller hang is observed with ATA pass through command followed =
by a
        pended command that fails.
        At the completion of pass through command, PL starts the pended IOs=
 if
any.
        If the pended IO is failed due to invalid CDB, then immediately
        completion function is called causing recursion. This causes contro=
ller
        hang.
    Steps To Reproduce:
        Install the FreeNAS and try to create ZFS pool(Storage -> Pools -> =
Add)
        of the direct attached SSDs.
        - ATA command completes when there are IOs in pendlist.
        - The pended IO has invalid CDB

I have installed the firmware, and am keeping an eye on it. My system is
lightly loaded though, so my sightings of the issue have been rather spurio=
us.

--=20
You are receiving this mail because:
You are the assignee for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-224496-227-UqiOaxnwlr>