Date: Sun, 24 Oct 2010 20:55:13 +0300 From: Alexander Motin <mav@FreeBSD.org> To: Garrett Cooper <gcooper@FreeBSD.org> Cc: Bruce Cran <bruce@cran.org.uk>, svn-src-head@freebsd.org, svn-src-all@freebsd.org, src-committers@freebsd.org Subject: Re: svn commit: r214279 - in head: share/man/man4 sys/cam sys/cam/ata sys/kern Message-ID: <4CC47301.9020609@FreeBSD.org> In-Reply-To: <AANLkTikj4Uo6haewmHcgXeONt5Dh6uQ0OsH6KuVwFGTx@mail.gmail.com> References: <201010241631.o9OGVw1C033817@svn.freebsd.org> <201010241735.32876.bruce@cran.org.uk> <4CC466D5.6080306@FreeBSD.org> <AANLkTikj4Uo6haewmHcgXeONt5Dh6uQ0OsH6KuVwFGTx@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Garrett Cooper wrote: > On Sun, Oct 24, 2010 at 10:03 AM, Alexander Motin <mav@freebsd.org> wrote: >> Bruce Cran wrote: >>> On Sunday 24 October 2010 17:31:58 Bruce Cran wrote: >>> >>>> Log: >>>> Mostly revert r203420, and add similar functionality into ada(4) since >>>> the existing code caused problems with some SCSI controllers. >> Proper way would be IMHO to fix polling in aac driver. >> >>>> A new sysctl kern.cam.ada.spindown_shutdown has been added that controls >>>> whether or not to spin-down disks when shutting down. >>>> Spinning down the disks unloads/parks the heads - this is >>>> much better than removing power when the disk is still >>>> spinning because otherwise an Emergency Unload occurs which may cause >>>> damage to the actuator. >>> The FLUSH CACHE + STANDBY IMMEDIATE commands are issued (instead of just >>> SLEEP) following the procedure documented in Fujitsu's MHW series product >>> documentation under section 1.10.1, "Recommended power-off sequence". >> Not instead of "just SLEEP", but instead of FLUSH CACHE (by respective >> peripheral driver) + SLEEP (by xpt). It should probably be the same. >> >> Just for the note, SCSI specification states that STOP automatically >> implies FLUSH CACHE. ATA - doesn't. > > I could be wrong, but I think the ANSI ATA<->SCSI spec states > otherwise (taken from ANSI INCITS 431-2007 page 62 -- see steps 1-4): > > 9.11.3 START STOP UNIT START bit LOEJ bit combinations > The SATL shall perform the actions shown in table 40 in response to a > START STOP UNIT command. > Table 40 — Definition of START and LOEJ bits in the START STOP UNIT CDB > START LOEJ Definition > 0 0 The SATL shall: > 1) If the IMMED bit is set to one, then return GOOD status; > 2) Issue an ATA flush command (see 3.1.11) to the ATA device; > 3) If the ATA flush command completes with any error, then process ending status > according to the IMMED bit (see 9.11.2) with the additional sense code set to > COMMAND SEQUENCE ERROR; > 4) If the ATA flush command completes without error, then issue an ATA STANDBY > IMMEDIATE command to the ATA Sector Count set to zero; > 5) If the ATA STANDBY IMMEDIATE command completes with any error, then > process ending status according to the IMMED bit (see 9.11.2) with the > additional > sense code set to COMMAND SEQUENCE ERROR; and > 6) If the ATA STANDBY IMMEDIATE command completes without error and the > IMMED bit is set to zero, then return GOOD status (see 9.11.2) a. > > Not sure about the ATA spec standalone... might be present in the ATA8 spec. As I understand, this just confirms what I have said: SCSI STOP == ATA FLUSH CACHE + STANDBY IMMEDIATE. So SCSI STOP implies flush, while ATA STANDBY IMMEDIATE doesn't. -- Alexander Motin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4CC47301.9020609>