Date: Sun, 15 Apr 2007 10:28:23 -0700 From: Nate Lawson <nate@root.org> To: Scott Long <scottl@FreeBSD.org> Cc: cvs-src@freebsd.org, src-committers@freebsd.org, cvs-all@freebsd.org Subject: Re: cvs commit: src/sys/cam cam_ccb.h cam_periph.c cam_periph.h cam_sim.c cam_sim.h cam_xpt.c cam_xpt.h cam_xpt_periph.h src/sys/cam/scsi scsi_cd.c scsi_ch.c scsi_da.c scsi_low.c scsi_pass.c scsi_pt.c scsi_sa.c scsi_ses.c scsi_sg.c ... Message-ID: <462260B7.8090602@root.org> In-Reply-To: <20070415084926.C77C816A521@hub.freebsd.org> References: <20070415084926.C77C816A521@hub.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Scott Long wrote:
> scottl 2007-04-15 08:49:19 UTC
>
> FreeBSD src repository
>
> Modified files:
> sys/cam cam_ccb.h cam_periph.c cam_periph.h
> cam_sim.c cam_sim.h cam_xpt.c cam_xpt.h
> cam_xpt_periph.h
> sys/cam/scsi scsi_cd.c scsi_ch.c scsi_da.c scsi_low.c
> scsi_pass.c scsi_pt.c scsi_sa.c
> scsi_ses.c scsi_sg.c scsi_targ_bh.c
> scsi_target.c
> sys/dev/aac aac_cam.c
> sys/dev/advansys advansys.c adwcam.c
> sys/dev/aha aha.c
> sys/dev/ahb ahb.c
> sys/dev/aic aic.c
> sys/dev/aic7xxx aic79xx_osm.c aic7xxx_osm.c
> sys/dev/amd amd.c
> sys/dev/amr amr_cam.c
> sys/dev/arcmsr arcmsr.c
> sys/dev/asr asr.c
> sys/dev/ata atapi-cam.c
> sys/dev/buslogic bt.c
> sys/dev/ciss ciss.c
> sys/dev/dpt dpt_scsi.c
> sys/dev/esp ncr53c9x.c
> sys/dev/firewire sbp.c sbp_targ.c
> sys/dev/hptmv entry.c
> sys/dev/iir iir.c
> sys/dev/isp isp_freebsd.c
> sys/dev/mly mly.c
> sys/dev/mpt mpt_cam.c
> sys/dev/ppbus vpo.c
> sys/dev/rr232x osm_bsd.c
> sys/dev/sym sym_hipd.c
> sys/dev/trm trm.c
> sys/dev/twa tw_osl_cam.c
> sys/dev/usb umass.c
> sys/dev/wds wd7000.c
> sys/pci ncr.c
> Log:
> Remove Giant from CAM. Drivers (SIMs) now register a mutex that CAM will
> use to synchornize and protect all data objects that are used for that
> SIM. Drivers that are not yet MPSAFE register Giant and operate as
> usual. RIght now, no drivers are MPSAFE, though a few will be changed
> in the coming week as this work settles down.
>
> The driver API has changed, so all CAM drivers will need to be recompiled.
> The userland API has not changed, so tools like camcontrol do not need to
> be recompiled.
>
> Revision Changes Path
> 1.35 +6 -3 src/sys/cam/cam_ccb.h
> 1.65 +86 -45 src/sys/cam/cam_periph.c
> 1.17 +5 -1 src/sys/cam/cam_periph.h
> 1.10 +32 -27 src/sys/cam/cam_sim.c
> 1.7 +19 -2 src/sys/cam/cam_sim.h
> 1.176 +447 -319 src/sys/cam/cam_xpt.c
> 1.8 +6 -0 src/sys/cam/cam_xpt.h
> 1.6 +2 -2 src/sys/cam/cam_xpt_periph.h
> 1.98 +244 -194 src/sys/cam/scsi/scsi_cd.c
> 1.44 +38 -31 src/sys/cam/scsi/scsi_ch.c
> 1.204 +85 -98 src/sys/cam/scsi/scsi_da.c
> 1.27 +2 -2 src/sys/cam/scsi/scsi_low.c
> 1.46 +32 -26 src/sys/cam/scsi/scsi_pass.c
> 1.45 +19 -76 src/sys/cam/scsi/scsi_pt.c
> 1.108 +47 -71 src/sys/cam/scsi/scsi_sa.c
> 1.34 +36 -17 src/sys/cam/scsi/scsi_ses.c
> 1.5 +38 -19 src/sys/cam/scsi/scsi_sg.c
> 1.24 +2 -8 src/sys/cam/scsi/scsi_targ_bh.c
> 1.73 +64 -16 src/sys/cam/scsi/scsi_target.c
> 1.27 +1 -1 src/sys/dev/aac/aac_cam.c
> 1.34 +1 -1 src/sys/dev/advansys/advansys.c
> 1.27 +1 -1 src/sys/dev/advansys/adwcam.c
> 1.63 +2 -2 src/sys/dev/aha/aha.c
> 1.41 +1 -1 src/sys/dev/ahb/ahb.c
> 1.26 +4 -1 src/sys/dev/aic/aic.c
> 1.27 +1 -1 src/sys/dev/aic7xxx/aic79xx_osm.c
> 1.48 +3 -2 src/sys/dev/aic7xxx/aic7xxx_osm.c
> 1.35 +3 -2 src/sys/dev/amd/amd.c
> 1.21 +1 -0 src/sys/dev/amr/amr_cam.c
> 1.22 +3 -2 src/sys/dev/arcmsr/arcmsr.c
> 1.84 +2 -1 src/sys/dev/asr/asr.c
> 1.51 +1 -1 src/sys/dev/ata/atapi-cam.c
> 1.49 +1 -1 src/sys/dev/buslogic/bt.c
> 1.76 +2 -2 src/sys/dev/ciss/ciss.c
> 1.56 +2 -1 src/sys/dev/dpt/dpt_scsi.c
> 1.16 +1 -1 src/sys/dev/esp/ncr53c9x.c
> 1.90 +1 -0 src/sys/dev/firewire/sbp.c
> 1.11 +1 -1 src/sys/dev/firewire/sbp_targ.c
> 1.16 +2 -1 src/sys/dev/hptmv/entry.c
> 1.18 +2 -1 src/sys/dev/iir/iir.c
> 1.138 +4 -3 src/sys/dev/isp/isp_freebsd.c
> 1.44 +2 -0 src/sys/dev/mly/mly.c
> 1.54 +7 -5 src/sys/dev/mpt/mpt_cam.c
> 1.35 +3 -1 src/sys/dev/ppbus/vpo.c
> 1.5 +2 -1 src/sys/dev/rr232x/osm_bsd.c
> 1.64 +1 -1 src/sys/dev/sym/sym_hipd.c
> 1.32 +1 -0 src/sys/dev/trm/trm.c
> 1.9 +1 -1 src/sys/dev/twa/tw_osl_cam.c
> 1.143 +3 -0 src/sys/dev/usb/umass.c
> 1.13 +2 -1 src/sys/dev/wds/wd7000.c
> 1.194 +3 -1 src/sys/pci/ncr.c
>
Amazing work, I owe you a beer for following through where none of us
did. I see you went with the SIM lock approach, which seems reasonable.
On the targ diff:
@@ -229,44 +252,56 @@
{
struct ioc_enable_lun *new_lun;
struct cam_path *path;
+ struct cam_sim *sim;
new_lun = (struct ioc_enable_lun *)addr;
- status = xpt_create_path(&path, /*periph*/NULL,
- new_lun->path_id,
- new_lun->target_id,
- new_lun->lun_id);
+ status = xpt_create_path_unlocked(&path, /*periph*/NULL,
+ new_lun->path_id,
+ new_lun->target_id,
+ new_lun->lun_id);
if (status != CAM_REQ_CMP) {
printf("Couldn't create path, status %#x\n", status);
break;
}
+ sim = xpt_path_sim(path);
+ mtx_lock(sim->mtx);
status = targenable(softc, path, new_lun->grp6_len,
new_lun->grp7_len);
xpt_free_path(path);
+ mtx_unlock(sim->mtx);
^^^^^^^^^^^^^^^^^^^^^
break;
}
I think the unlock can come before freeing the path, can't it?
--
Nate
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?462260B7.8090602>