Index: stable/10/sbin/devd/devd.conf.5 =================================================================== --- stable/10/sbin/devd/devd.conf.5 (revision 300213) +++ stable/10/sbin/devd/devd.conf.5 (revision 300214) @@ -1,642 +1,644 @@ .\" .\" Copyright (c) 2002 M. Warner Losh .\" All rights reserved. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions .\" are met: .\" 1. Redistributions of source code must retain the above copyright .\" notice, this list of conditions and the following disclaimer. .\" 2. The name of the author may not be used to endorse or promote products .\" derived from this software without specific prior written permission. .\" .\" THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND .\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE .\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL .\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS .\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) .\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT .\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY .\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF .\" SUCH DAMAGE. .\" .\" $FreeBSD$ .\" .\" The section on comments was taken from named.conf.5, which has the .\" following copyright: .\" Copyright (c) 1999-2000 by Internet Software Consortium .\" .\" Permission to use, copy, modify, and distribute this software for any .\" purpose with or without fee is hereby granted, provided that the above .\" copyright notice and this permission notice appear in all copies. .\" .\" THE SOFTWARE IS PROVIDED "AS IS" AND INTERNET SOFTWARE CONSORTIUM DISCLAIMS .\" ALL WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES .\" OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL INTERNET SOFTWARE .\" CONSORTIUM BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL .\" DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR .\" PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS .\" ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS .\" SOFTWARE. .\" .Dd April 14, 2016 .Dt DEVD.CONF 5 .Os .Sh NAME .Nm devd.conf .Nd configuration file for .Xr devd 8 .Sh DESCRIPTION .Ss General Syntax A .Xr devd 8 configuration consists of two general features, statements and comments. All statements end with a semicolon. Many statements can contain substatements, which are also terminated with a semicolon. .Pp The following statements are supported: .Bl -tag -width ".Ic options" .It Ic attach Specifies various matching criteria and actions to perform when a newly attached device matches said criteria. .It Ic detach Specifies various matching criteria and actions to perform when a newly detached device matches said criteria. .It Ic nomatch Specifies various matching criteria and actions to perform when no device driver currently loaded in the kernel claims a (new) device. .It Ic notify Specifies various matching criteria and actions to perform when the kernel sends an event notification to userland. .It Ic options Specifies various options and parameters for the operation of .Xr devd 8 . .El .Pp Statements may occur in any order in the configuration file, and may be repeated as often as required. Further details on the syntax and meaning of each statement and their substatements are explained below. .Pp Each statement, except .Ic options has a priority (an arbitrary number) associated with it, where .Ql 0 is defined as the lowest priority. If two statements match the same event, only the action of the statement with highest priority will be executed. In this way generic statements can be overridden for devices or notifications that require special attention. .Pp The general syntax of a statement is: .Bd -literal -offset indent statement priority { substatement "value"; ... substatement "value"; }; .Ed .Ss Sub-statements The following sub-statements are supported within the .Ic options statement. .Bl -tag -width ".Ic directory" .It Ic directory Qq Ar /some/path ; Adds the given directory to the list of directories from which .Xr devd 8 will read all files named "*.conf" as further configuration files. Any number of .Ic directory statements can be used. .It Ic pid-file Qq Pa /var/run/devd.pid ; Specifies PID file. .It Ic set Ar regexp-name Qq Ar (some|regexp) ; Creates a regular expression and assigns it to the variable .Ar regexp-name . The variable is available throughout the rest of the configuration file. If the string begins with .Ql \&! , it matches if the regular expression formed by the rest of the string does not match. All regular expressions have an implicit .Ql ^$ around them. .El .Pp The following sub-statements are supported within the .Ic attach and .Ic detach statements. .Bl -tag -width ".Ic directory" .It Ic action Qq Ar command ; Command to execute upon a successful match. Example .Dq Li "/etc/pccard_ether $device-name start" . .It Ic class Qq Ar string ; This is shorthand for .Dq Ic match Qo Li class Qc Qq Ar string . .It Ic device-name Qq string ; This is shorthand for .Dq Ic match Qo Li device-name Qc Qq Ar string . This matches a device named .Ar string , which is allowed to be a regular expression or a variable previously created containing a regular expression. The .Dq Li device-name variable is available for later use with the .Ic action statement. .It Ic match Qo Ar variable Qc Qq Ar value ; Matches the content of .Ar value against .Ar variable ; the content of .Ar value may be a regular expression. Not required during .Ic attach nor .Ic detach events since the .Ic device-name statement takes care of all device matching. For a partial list of variables, see below. .It Ic media-type Qq Ar string ; For network devices, .Ic media-type will match devices that have the given media type. Valid media types are: .Dq Li Ethernet , .Dq Li Tokenring , .Dq Li FDDI , .Dq Li 802.11 , and .Dq Li ATM . .It Ic subdevice Qq Ar string ; This is shorthand for .Dq Ic match Qo Li subdevice Qc Qq Ar string . .El .Pp The following sub-statements are supported within the .Ic nomatch statement. .Bl -tag -width ".Ic directory" .It Ic action Qq Ar command ; Same as above. .It Ic match Qo Ar variable Qc Qq Ar value ; Matches the content of .Ar value against .Ar variable ; the content of .Ar value may be a regular expression. For a partial list of variables, see below. .El .Pp The following sub-statements are supported within the .Ic notify statement. The .Dq Li notify variable is available inside this statement and contains, a value, depending on which system and subsystem that delivered the event. .Bl -tag -width ".Ic directory" .It Ic action Qq Ar command ; Command to execute upon a successful match. Example .Dq Li "/etc/rc.d/power_profile $notify" . .It Ic match Qo Ar system | subsystem | type | notify Qc Qq Ar value ; Any number of .Ic match statements can exist within a .Ic notify statement; .Ar value can be either a fixed string or a regular expression. Below is a list of available systems, subsystems, and types. .It Ic media-type Qq Ar string ; See above. .El .Ss Variables that can be used with the match statement A partial list of variables and their possible values that can be used together with the .Ic match statement. .Pp .Bl -tag -width ".Li manufacturer" -compact .It Ic Variable .Ic Description .It Li bus Device name of parent bus. .It Li cdev Device node path if one is created by the .Xr devfs 5 filesystem. .It Li cisproduct CIS-product. .It Li cisvendor CIS-vendor. .It Li class Device class. .It Li device Device ID. .It Li devclass Device Class (USB) .It Li devsubclass Device Sub-class (USB) .It Li device-name Name of attached/detached device. .It Li endpoints Endpoint count (USB) .It Li function Card functions. .It Li interface Interface ID (USB) .It Li intclass Interface Class (USB) .It Li intprotocol Interface Protocol (USB) .It Li intsubclass Interface Sub-class (USB) .It Li jail Jail name for the process triggering the rule (RCTL) .It Li manufacturer Manufacturer ID (pccard). .It Li mode Peripheral mode (USB) .It Li notify Match the value of the .Dq Li notify variable. .It Li parent Parent device .It Li pid PID of the process triggering the rule (RCTL) .It Li port Hub port number (USB) .It Li product Product ID (pccard/USB). .It Li release Hardware revision (USB) .It Li ruid Real UID of the process triggering the rule (RCTL) .It Li rule Rule (RCTL) .It Li sernum Serial Number (USB). .It Li slot Card slot. .It Li subvendor Sub-vendor ID. .It Li subdevice Sub-device ID. .It Li subsystem Matches a subsystem of a system, see below. .It Li system Matches a system type, see below. .It Li type Type of notification, see below. .It Li vendor Vendor ID. .El .Ss Notify matching A partial list of systems, subsystems, and types used within the .Ic notify mechanism. .Pp .Bl -tag -width ".Li coretemp" -compact .It Sy System .It Li ACPI Events related to the ACPI subsystem. .Bl -tag -width ".Sy Subsystem" -compact .It Sy Subsystem .It Li ACAD AC line state ($notify=0x00 is offline, 0x01 is online). .It Li Button Button state ($notify=0x00 is power, 0x01 is sleep). .It Li CMBAT Battery events. .It Li Lid Lid state ($notify=0x00 is closed, 0x01 is open). .It Li PROCESSOR Processor state/configuration ($notify=0x81 is a change in available Cx states). .It Li Resume Resume notification. .It Li Suspend Suspend notification. .It Li Thermal Thermal zone events. .El .Pp .It Li CARP Events related to the .Xr carp 8 protocol. .Bl -tag -width ".Sy Subsystem" -compact .It Sy Subsystem .It Ar vhid@interface The .Dq subsystem contains the actual CARP vhid and the name of the network interface on which the event took place. .Bl -tag -width ".Li MASTER" -compact .It Sy Type .It Li MASTER Node become the master for a virtual host. .It Li BACKUP Node become the backup for a virtual host. .El .El .Pp .It Li IFNET Events related to the network subsystem. .Bl -tag -width ".Sy Subsystem" -compact .It Sy Subsystem .It Ar interface The .Dq subsystem is the actual name of the network interface on which the event took place. .Bl -tag -width ".Li LINK_DOWN" -compact .It Sy Type .It Li LINK_UP Carrier status changed to UP. .It Li LINK_DOWN Carrier status changed to DOWN. .It Li ATTACH The network interface is attached to the system. .It Li DETACH The network interface is detached from the system. .El .El .Pp .It Li DEVFS Events related to the .Xr devfs 5 filesystem. .Bl -tag -width ".Sy Subsystem" -compact .It Sy Subsystem .It Li CDEV .Bl -tag -width ".Li DESTROY" -compact .It Sy Type .It Li CREATE The .Xr devfs 5 node is created. .It Li DESTROY The .Xr devfs 5 node is destroyed. .El .El .Pp .It Li GEOM Events related to the .Xr geom 4 framework. The difference compared to .Li DEVFS is that .Li GEOM only includes disk-like devices. .Bl -tag -width ".Sy Subsystem" -compact .It Sy Subsystem .It Li DEV .Bl -tag -width ".Li MEDIACHANGE" -compact .It Sy Type .It Li CREATE A .Xr geom 4 device node is created. .It Li DESTROY A .Xr geom 4 device node is destroyed. +.It Li GEOM::physpath +The physical path of a device has changed. .It Li MEDIACHANGE Physical media has changed. .El .El .Pp .It Li RCTL Events related to the .Xr rctl 8 framework. .Bl -tag -width ".Sy Subsystem" -compact .It Sy Subsystem .It Li rule .Bl -tag -width ".Li matched" -compact .It Sy Type .It Li matched A rule with action specified as "devctl" was triggered. .El .El .Pp .It Li USB Events related to the USB subsystem. .Bl -tag -width ".Sy Subsystem" -compact .It Sy Subsystem .It Li DEVICE .Bl -tag -width ".Li DETACH" -compact .It Sy Type .It Li ATTACH USB device is attached to the system. .It Li DETACH USB device is detached from the system. .El .It Li INTERFACE .Bl -tag -width ".Li DETACH" -compact .It Sy Type .It Li ATTACH USB interface is attached to a device. .It Li DETACH USB interface is detached from a device. .El .El .Pp .It Li coretemp Events related to the .Xr coretemp 4 device. .Bl -tag -width ".Sy Subsystem" -compact .It Sy Subsystem .It Li Thermal Notification that the CPU core has reached critical temperature. .Bl -tag -width ".Ar temperature" -compact .It Sy Type .It Ar temperature String containing the temperature of the core that has become too hot. .El .El .Pp .It Li kern Events related to the kernel. .Bl -tag -width ".Sy Subsystem" -compact .It Sy Subsystem .It Li power Information about the state of the system. .Bl -tag -width ".li resume" -compact .It Sy Type .It Li resume Notification that the system has woken from the suspended state. .El .El .Pp .It Li CAM Events related to the .Xr cam 4 system. .Bl -tag -width ".Sy Subsystem" -compact .It Sy Subsystem .It Li periph Events related to peripheral devices. .Bl -tag -width ".li timeout" -compact .It Sy Type .It Li error Generic errors. .It Li timeout Command timeouts. .El .El .El .Pp A link state change to UP on the interface .Dq Li fxp0 would result in the following notify event: .Bd -literal -offset indent system=IFNET, subsystem=fxp0, type=LINK_UP .Ed .Pp An AC line state change to .Dq offline would result in the following event: .Bd -literal -offset indent system=ACPI, subsystem=ACAD, notify=0x00 .Ed .Ss Comments Comments may appear anywhere that whitespace may appear in a configuration file. To appeal to programmers of all kinds, they can be written in C, C++, or shell/Perl constructs. .Pp C-style comments start with the two characters .Ql /* (slash, star) and end with .Ql */ (star, slash). Because they are completely delimited with these characters, they can be used to comment only a portion of a line or to span multiple lines. .Pp C-style comments cannot be nested. For example, the following is not valid because the entire comment ends with the first .Ql */ : .Bd -literal -offset indent /* This is the start of a comment. This is still part of the comment. /* This is an incorrect attempt at nesting a comment. */ This is no longer in any comment. */ .Ed .Pp C++-style comments start with the two characters .Ql // (slash, slash) and continue to the end of the physical line. They cannot be continued across multiple physical lines; to have one logical comment span multiple lines, each line must use the .Ql // pair. For example: .Bd -literal -offset indent // This is the start of a comment. The next line // is a new comment, even though it is logically // part of the previous comment. .Ed .Sh FILES .Bl -tag -width ".Pa /etc/devd.conf" -compact .It Pa /etc/devd.conf The .Xr devd 8 configuration file. .El .Sh EXAMPLES .Bd -literal # # This will catch link down events on the interfaces fxp0 and ath0 # notify 0 { match "system" "IFNET"; match "subsystem" "(fxp0|ath0)"; match "type" "LINK_DOWN"; action "logger $subsystem is DOWN"; }; # # Match lid open/close events # These can be combined to a single event, by passing the # value of $notify to the external script. # notify 0 { match "system" "ACPI"; match "subsystem" "Lid"; match "notify" "0x00"; action "logger Lid closed, we can sleep now!"; }; notify 0 { match "system" "ACPI"; match "subsystem" "Lid"; match "notify" "0x01"; action "logger Lid opened, the sleeper must awaken!"; }; # # Match a USB device type # notify 0 { match "system" "USB"; match "subsystem" "INTERFACE"; match "type" "ATTACH"; match "intclass" "0x0e"; action "logger USB video device attached"; }; # # Try to configure ath and wi devices with pccard_ether # as they are attached. # attach 0 { device-name "(ath|wi)[0-9]+"; action "/etc/pccard_ether $device-name start"; }; # # Stop ath and wi devices as they are detached from # the system. # detach 0 { device-name "(ath|wi)[0-9]+"; action "/etc/pccard_ether $device-name stop"; }; .Ed .Pp The installed .Pa /etc/devd.conf has many additional examples. .Sh SEE ALSO .Xr coretemp 4 , .Xr devfs 5 , .Xr re_format 7 , .Xr devd 8 , .Xr cam 4 Index: stable/10/sys/geom/geom_disk.c =================================================================== --- stable/10/sys/geom/geom_disk.c (revision 300213) +++ stable/10/sys/geom/geom_disk.c (revision 300214) @@ -1,947 +1,952 @@ /*- * Copyright (c) 2002 Poul-Henning Kamp * Copyright (c) 2002 Networks Associates Technology, Inc. * All rights reserved. * * This software was developed for the FreeBSD Project by Poul-Henning Kamp * and NAI Labs, the Security Research Division of Network Associates, Inc. * under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the * DARPA CHATS research program. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. The names of the authors may not be used to endorse or promote * products derived from this software without specific prior written * permission. * * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ #include __FBSDID("$FreeBSD$"); #include "opt_geom.h" #include #include #include #include #include +#include #include #include #include #include #include #include #include #include #include #include #include #include #include struct g_disk_softc { struct mtx done_mtx; struct disk *dp; struct sysctl_ctx_list sysctl_ctx; struct sysctl_oid *sysctl_tree; char led[64]; uint32_t state; struct mtx start_mtx; }; static g_access_t g_disk_access; static g_start_t g_disk_start; static g_ioctl_t g_disk_ioctl; static g_dumpconf_t g_disk_dumpconf; static g_provgone_t g_disk_providergone; static struct g_class g_disk_class = { .name = G_DISK_CLASS_NAME, .version = G_VERSION, .start = g_disk_start, .access = g_disk_access, .ioctl = g_disk_ioctl, .providergone = g_disk_providergone, .dumpconf = g_disk_dumpconf, }; SYSCTL_DECL(_kern_geom); static SYSCTL_NODE(_kern_geom, OID_AUTO, disk, CTLFLAG_RW, 0, "GEOM_DISK stuff"); DECLARE_GEOM_CLASS(g_disk_class, g_disk); static void __inline g_disk_lock_giant(struct disk *dp) { if (dp->d_flags & DISKFLAG_NEEDSGIANT) mtx_lock(&Giant); } static void __inline g_disk_unlock_giant(struct disk *dp) { if (dp->d_flags & DISKFLAG_NEEDSGIANT) mtx_unlock(&Giant); } static int g_disk_access(struct g_provider *pp, int r, int w, int e) { struct disk *dp; struct g_disk_softc *sc; int error; g_trace(G_T_ACCESS, "g_disk_access(%s, %d, %d, %d)", pp->name, r, w, e); g_topology_assert(); sc = pp->private; if (sc == NULL || (dp = sc->dp) == NULL || dp->d_destroyed) { /* * Allow decreasing access count even if disk is not * avaliable anymore. */ if (r <= 0 && w <= 0 && e <= 0) return (0); return (ENXIO); } r += pp->acr; w += pp->acw; e += pp->ace; error = 0; if ((pp->acr + pp->acw + pp->ace) == 0 && (r + w + e) > 0) { if (dp->d_open != NULL) { g_disk_lock_giant(dp); error = dp->d_open(dp); if (bootverbose && error != 0) printf("Opened disk %s -> %d\n", pp->name, error); g_disk_unlock_giant(dp); if (error != 0) return (error); } pp->mediasize = dp->d_mediasize; pp->sectorsize = dp->d_sectorsize; if (dp->d_maxsize == 0) { printf("WARNING: Disk drive %s%d has no d_maxsize\n", dp->d_name, dp->d_unit); dp->d_maxsize = DFLTPHYS; } if (dp->d_delmaxsize == 0) { if (bootverbose && dp->d_flags & DISKFLAG_CANDELETE) { printf("WARNING: Disk drive %s%d has no " "d_delmaxsize\n", dp->d_name, dp->d_unit); } dp->d_delmaxsize = dp->d_maxsize; } pp->stripeoffset = dp->d_stripeoffset; pp->stripesize = dp->d_stripesize; dp->d_flags |= DISKFLAG_OPEN; } else if ((pp->acr + pp->acw + pp->ace) > 0 && (r + w + e) == 0) { if (dp->d_close != NULL) { g_disk_lock_giant(dp); error = dp->d_close(dp); if (error != 0) printf("Closed disk %s -> %d\n", pp->name, error); g_disk_unlock_giant(dp); } sc->state = G_STATE_ACTIVE; if (sc->led[0] != 0) led_set(sc->led, "0"); dp->d_flags &= ~DISKFLAG_OPEN; } return (error); } static void g_disk_kerneldump(struct bio *bp, struct disk *dp) { struct g_kerneldump *gkd; struct g_geom *gp; gkd = (struct g_kerneldump*)bp->bio_data; gp = bp->bio_to->geom; g_trace(G_T_TOPOLOGY, "g_disk_kerneldump(%s, %jd, %jd)", gp->name, (intmax_t)gkd->offset, (intmax_t)gkd->length); if (dp->d_dump == NULL) { g_io_deliver(bp, ENODEV); return; } gkd->di.dumper = dp->d_dump; gkd->di.priv = dp; gkd->di.blocksize = dp->d_sectorsize; gkd->di.maxiosize = dp->d_maxsize; gkd->di.mediaoffset = gkd->offset; if ((gkd->offset + gkd->length) > dp->d_mediasize) gkd->length = dp->d_mediasize - gkd->offset; gkd->di.mediasize = gkd->length; g_io_deliver(bp, 0); } static void g_disk_setstate(struct bio *bp, struct g_disk_softc *sc) { const char *cmd; memcpy(&sc->state, bp->bio_data, sizeof(sc->state)); if (sc->led[0] != 0) { switch (sc->state) { case G_STATE_FAILED: cmd = "1"; break; case G_STATE_REBUILD: cmd = "f5"; break; case G_STATE_RESYNC: cmd = "f1"; break; default: cmd = "0"; break; } led_set(sc->led, cmd); } g_io_deliver(bp, 0); } static void g_disk_done(struct bio *bp) { struct bintime now; struct bio *bp2; struct g_disk_softc *sc; /* See "notes" for why we need a mutex here */ /* XXX: will witness accept a mix of Giant/unGiant drivers here ? */ bp2 = bp->bio_parent; sc = bp2->bio_to->private; bp->bio_completed = bp->bio_length - bp->bio_resid; binuptime(&now); mtx_lock(&sc->done_mtx); if (bp2->bio_error == 0) bp2->bio_error = bp->bio_error; bp2->bio_completed += bp->bio_completed; if ((bp->bio_cmd & (BIO_READ|BIO_WRITE|BIO_DELETE|BIO_FLUSH)) != 0) devstat_end_transaction_bio_bt(sc->dp->d_devstat, bp, &now); bp2->bio_inbed++; if (bp2->bio_children == bp2->bio_inbed) { mtx_unlock(&sc->done_mtx); bp2->bio_resid = bp2->bio_bcount - bp2->bio_completed; g_io_deliver(bp2, bp2->bio_error); } else mtx_unlock(&sc->done_mtx); g_destroy_bio(bp); } static int g_disk_ioctl(struct g_provider *pp, u_long cmd, void * data, int fflag, struct thread *td) { struct disk *dp; struct g_disk_softc *sc; int error; sc = pp->private; dp = sc->dp; if (dp->d_ioctl == NULL) return (ENOIOCTL); g_disk_lock_giant(dp); error = dp->d_ioctl(dp, cmd, data, fflag, td); g_disk_unlock_giant(dp); return (error); } static off_t g_disk_maxsize(struct disk *dp, struct bio *bp) { if (bp->bio_cmd == BIO_DELETE) return (dp->d_delmaxsize); return (dp->d_maxsize); } static int g_disk_maxsegs(struct disk *dp, struct bio *bp) { return ((g_disk_maxsize(dp, bp) / PAGE_SIZE) + 1); } static void g_disk_advance(struct disk *dp, struct bio *bp, off_t off) { bp->bio_offset += off; bp->bio_length -= off; if ((bp->bio_flags & BIO_VLIST) != 0) { bus_dma_segment_t *seg, *end; seg = (bus_dma_segment_t *)bp->bio_data; end = (bus_dma_segment_t *)bp->bio_data + bp->bio_ma_n; off += bp->bio_ma_offset; while (off >= seg->ds_len) { KASSERT((seg != end), ("vlist request runs off the end")); off -= seg->ds_len; seg++; } bp->bio_ma_offset = off; bp->bio_ma_n = end - seg; bp->bio_data = (void *)seg; } else if ((bp->bio_flags & BIO_UNMAPPED) != 0) { bp->bio_ma += off / PAGE_SIZE; bp->bio_ma_offset += off; bp->bio_ma_offset %= PAGE_SIZE; bp->bio_ma_n -= off / PAGE_SIZE; } else { bp->bio_data += off; } } static void g_disk_seg_limit(bus_dma_segment_t *seg, off_t *poffset, off_t *plength, int *ppages) { uintptr_t seg_page_base; uintptr_t seg_page_end; off_t offset; off_t length; int seg_pages; offset = *poffset; length = *plength; if (length > seg->ds_len - offset) length = seg->ds_len - offset; seg_page_base = trunc_page(seg->ds_addr + offset); seg_page_end = round_page(seg->ds_addr + offset + length); seg_pages = (seg_page_end - seg_page_base) >> PAGE_SHIFT; if (seg_pages > *ppages) { seg_pages = *ppages; length = (seg_page_base + (seg_pages << PAGE_SHIFT)) - (seg->ds_addr + offset); } *poffset = 0; *plength -= length; *ppages -= seg_pages; } static off_t g_disk_vlist_limit(struct disk *dp, struct bio *bp, bus_dma_segment_t **pendseg) { bus_dma_segment_t *seg, *end; off_t residual; off_t offset; int pages; seg = (bus_dma_segment_t *)bp->bio_data; end = (bus_dma_segment_t *)bp->bio_data + bp->bio_ma_n; residual = bp->bio_length; offset = bp->bio_ma_offset; pages = g_disk_maxsegs(dp, bp); while (residual != 0 && pages != 0) { KASSERT((seg != end), ("vlist limit runs off the end")); g_disk_seg_limit(seg, &offset, &residual, &pages); seg++; } if (pendseg != NULL) *pendseg = seg; return (residual); } static bool g_disk_limit(struct disk *dp, struct bio *bp) { bool limited = false; off_t maxsz; maxsz = g_disk_maxsize(dp, bp); /* * XXX: If we have a stripesize we should really use it here. * Care should be taken in the delete case if this is done * as deletes can be very sensitive to size given how they * are processed. */ if (bp->bio_length > maxsz) { bp->bio_length = maxsz; limited = true; } if ((bp->bio_flags & BIO_VLIST) != 0) { bus_dma_segment_t *firstseg, *endseg; off_t residual; firstseg = (bus_dma_segment_t*)bp->bio_data; residual = g_disk_vlist_limit(dp, bp, &endseg); if (residual != 0) { bp->bio_ma_n = endseg - firstseg; bp->bio_length -= residual; limited = true; } } else if ((bp->bio_flags & BIO_UNMAPPED) != 0) { bp->bio_ma_n = howmany(bp->bio_ma_offset + bp->bio_length, PAGE_SIZE); } return (limited); } static void g_disk_start(struct bio *bp) { struct bio *bp2, *bp3; struct disk *dp; struct g_disk_softc *sc; int error; off_t off; sc = bp->bio_to->private; if (sc == NULL || (dp = sc->dp) == NULL || dp->d_destroyed) { g_io_deliver(bp, ENXIO); return; } error = EJUSTRETURN; switch(bp->bio_cmd) { case BIO_DELETE: if (!(dp->d_flags & DISKFLAG_CANDELETE)) { error = EOPNOTSUPP; break; } /* fall-through */ case BIO_READ: case BIO_WRITE: KASSERT((dp->d_flags & DISKFLAG_UNMAPPED_BIO) != 0 || (bp->bio_flags & BIO_UNMAPPED) == 0, ("unmapped bio not supported by disk %s", dp->d_name)); off = 0; bp3 = NULL; bp2 = g_clone_bio(bp); if (bp2 == NULL) { error = ENOMEM; break; } for (;;) { if (g_disk_limit(dp, bp2)) { off += bp2->bio_length; /* * To avoid a race, we need to grab the next bio * before we schedule this one. See "notes". */ bp3 = g_clone_bio(bp); if (bp3 == NULL) bp->bio_error = ENOMEM; } bp2->bio_done = g_disk_done; bp2->bio_pblkno = bp2->bio_offset / dp->d_sectorsize; bp2->bio_bcount = bp2->bio_length; bp2->bio_disk = dp; mtx_lock(&sc->start_mtx); devstat_start_transaction_bio(dp->d_devstat, bp2); mtx_unlock(&sc->start_mtx); g_disk_lock_giant(dp); dp->d_strategy(bp2); g_disk_unlock_giant(dp); if (bp3 == NULL) break; bp2 = bp3; bp3 = NULL; g_disk_advance(dp, bp2, off); } break; case BIO_GETATTR: /* Give the driver a chance to override */ if (dp->d_getattr != NULL) { if (bp->bio_disk == NULL) bp->bio_disk = dp; error = dp->d_getattr(bp); if (error != -1) break; error = EJUSTRETURN; } if (g_handleattr_int(bp, "GEOM::candelete", (dp->d_flags & DISKFLAG_CANDELETE) != 0)) break; else if (g_handleattr_int(bp, "GEOM::fwsectors", dp->d_fwsectors)) break; else if (g_handleattr_int(bp, "GEOM::fwheads", dp->d_fwheads)) break; else if (g_handleattr_off_t(bp, "GEOM::frontstuff", 0)) break; else if (g_handleattr_str(bp, "GEOM::ident", dp->d_ident)) break; else if (g_handleattr_uint16_t(bp, "GEOM::hba_vendor", dp->d_hba_vendor)) break; else if (g_handleattr_uint16_t(bp, "GEOM::hba_device", dp->d_hba_device)) break; else if (g_handleattr_uint16_t(bp, "GEOM::hba_subvendor", dp->d_hba_subvendor)) break; else if (g_handleattr_uint16_t(bp, "GEOM::hba_subdevice", dp->d_hba_subdevice)) break; else if (!strcmp(bp->bio_attribute, "GEOM::kerneldump")) g_disk_kerneldump(bp, dp); else if (!strcmp(bp->bio_attribute, "GEOM::setstate")) g_disk_setstate(bp, sc); else if (!strcmp(bp->bio_attribute, "GEOM::rotation_rate")) { uint64_t v; if ((dp->d_flags & DISKFLAG_LACKS_ROTRATE) == 0) v = dp->d_rotation_rate; else v = 0; /* rate unknown */ g_handleattr_uint16_t(bp, "GEOM::rotation_rate", v); break; } else error = ENOIOCTL; break; case BIO_FLUSH: g_trace(G_T_BIO, "g_disk_flushcache(%s)", bp->bio_to->name); if (!(dp->d_flags & DISKFLAG_CANFLUSHCACHE)) { error = EOPNOTSUPP; break; } bp2 = g_clone_bio(bp); if (bp2 == NULL) { g_io_deliver(bp, ENOMEM); return; } bp2->bio_done = g_disk_done; bp2->bio_disk = dp; mtx_lock(&sc->start_mtx); devstat_start_transaction_bio(dp->d_devstat, bp2); mtx_unlock(&sc->start_mtx); g_disk_lock_giant(dp); dp->d_strategy(bp2); g_disk_unlock_giant(dp); break; default: error = EOPNOTSUPP; break; } if (error != EJUSTRETURN) g_io_deliver(bp, error); return; } static void g_disk_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) { struct bio *bp; struct disk *dp; struct g_disk_softc *sc; char *buf; int res = 0; sc = gp->softc; if (sc == NULL || (dp = sc->dp) == NULL) return; if (indent == NULL) { sbuf_printf(sb, " hd %u", dp->d_fwheads); sbuf_printf(sb, " sc %u", dp->d_fwsectors); return; } if (pp != NULL) { sbuf_printf(sb, "%s%u\n", indent, dp->d_fwheads); sbuf_printf(sb, "%s%u\n", indent, dp->d_fwsectors); /* * "rotationrate" is a little complicated, because the value * returned by the drive might not be the RPM; 0 and 1 are * special cases, and there's also a valid range. */ sbuf_printf(sb, "%s", indent); if (dp->d_rotation_rate == 0) /* Old drives don't */ sbuf_printf(sb, "unknown"); /* report RPM. */ else if (dp->d_rotation_rate == 1) /* Since 0 is used */ sbuf_printf(sb, "0"); /* above, SSDs use 1. */ else if ((dp->d_rotation_rate >= 0x041) && (dp->d_rotation_rate <= 0xfffe)) sbuf_printf(sb, "%u", dp->d_rotation_rate); else sbuf_printf(sb, "invalid"); sbuf_printf(sb, "\n"); if (dp->d_getattr != NULL) { buf = g_malloc(DISK_IDENT_SIZE, M_WAITOK); bp = g_alloc_bio(); bp->bio_disk = dp; bp->bio_attribute = "GEOM::ident"; bp->bio_length = DISK_IDENT_SIZE; bp->bio_data = buf; res = dp->d_getattr(bp); sbuf_printf(sb, "%s", indent); g_conf_printf_escaped(sb, "%s", res == 0 ? buf: dp->d_ident); sbuf_printf(sb, "\n"); bp->bio_attribute = "GEOM::lunid"; bp->bio_length = DISK_IDENT_SIZE; bp->bio_data = buf; if (dp->d_getattr(bp) == 0) { sbuf_printf(sb, "%s", indent); g_conf_printf_escaped(sb, "%s", buf); sbuf_printf(sb, "\n"); } bp->bio_attribute = "GEOM::lunname"; bp->bio_length = DISK_IDENT_SIZE; bp->bio_data = buf; if (dp->d_getattr(bp) == 0) { sbuf_printf(sb, "%s", indent); g_conf_printf_escaped(sb, "%s", buf); sbuf_printf(sb, "\n"); } g_destroy_bio(bp); g_free(buf); } else { sbuf_printf(sb, "%s", indent); g_conf_printf_escaped(sb, "%s", dp->d_ident); sbuf_printf(sb, "\n"); } sbuf_printf(sb, "%s", indent); g_conf_printf_escaped(sb, "%s", dp->d_descr); sbuf_printf(sb, "\n"); } } static void g_disk_resize(void *ptr, int flag) { struct disk *dp; struct g_geom *gp; struct g_provider *pp; if (flag == EV_CANCEL) return; g_topology_assert(); dp = ptr; gp = dp->d_geom; if (dp->d_destroyed || gp == NULL) return; LIST_FOREACH(pp, &gp->provider, provider) { if (pp->sectorsize != 0 && pp->sectorsize != dp->d_sectorsize) g_wither_provider(pp, ENXIO); else g_resize_provider(pp, dp->d_mediasize); } } static void g_disk_create(void *arg, int flag) { struct g_geom *gp; struct g_provider *pp; struct disk *dp; struct g_disk_softc *sc; char tmpstr[80]; if (flag == EV_CANCEL) return; g_topology_assert(); dp = arg; sc = g_malloc(sizeof(*sc), M_WAITOK | M_ZERO); mtx_init(&sc->start_mtx, "g_disk_start", NULL, MTX_DEF); mtx_init(&sc->done_mtx, "g_disk_done", NULL, MTX_DEF); sc->dp = dp; gp = g_new_geomf(&g_disk_class, "%s%d", dp->d_name, dp->d_unit); gp->softc = sc; pp = g_new_providerf(gp, "%s", gp->name); devstat_remove_entry(pp->stat); pp->stat = NULL; dp->d_devstat->id = pp; pp->mediasize = dp->d_mediasize; pp->sectorsize = dp->d_sectorsize; pp->stripeoffset = dp->d_stripeoffset; pp->stripesize = dp->d_stripesize; if ((dp->d_flags & DISKFLAG_UNMAPPED_BIO) != 0) pp->flags |= G_PF_ACCEPT_UNMAPPED; if ((dp->d_flags & DISKFLAG_DIRECT_COMPLETION) != 0) pp->flags |= G_PF_DIRECT_SEND; pp->flags |= G_PF_DIRECT_RECEIVE; if (bootverbose) printf("GEOM: new disk %s\n", gp->name); sysctl_ctx_init(&sc->sysctl_ctx); snprintf(tmpstr, sizeof(tmpstr), "GEOM disk %s", gp->name); sc->sysctl_tree = SYSCTL_ADD_NODE(&sc->sysctl_ctx, SYSCTL_STATIC_CHILDREN(_kern_geom_disk), OID_AUTO, gp->name, CTLFLAG_RD, 0, tmpstr); if (sc->sysctl_tree != NULL) { snprintf(tmpstr, sizeof(tmpstr), "kern.geom.disk.%s.led", gp->name); TUNABLE_STR_FETCH(tmpstr, sc->led, sizeof(sc->led)); SYSCTL_ADD_STRING(&sc->sysctl_ctx, SYSCTL_CHILDREN(sc->sysctl_tree), OID_AUTO, "led", CTLFLAG_RW | CTLFLAG_TUN, sc->led, sizeof(sc->led), "LED name"); } pp->private = sc; dp->d_geom = gp; g_error_provider(pp, 0); } /* * We get this callback after all of the consumers have gone away, and just * before the provider is freed. If the disk driver provided a d_gone * callback, let them know that it is okay to free resources -- they won't * be getting any more accesses from GEOM. */ static void g_disk_providergone(struct g_provider *pp) { struct disk *dp; struct g_disk_softc *sc; sc = (struct g_disk_softc *)pp->private; dp = sc->dp; if (dp != NULL && dp->d_gone != NULL) dp->d_gone(dp); if (sc->sysctl_tree != NULL) { sysctl_ctx_free(&sc->sysctl_ctx); sc->sysctl_tree = NULL; } if (sc->led[0] != 0) { led_set(sc->led, "0"); sc->led[0] = 0; } pp->private = NULL; pp->geom->softc = NULL; mtx_destroy(&sc->done_mtx); mtx_destroy(&sc->start_mtx); g_free(sc); } static void g_disk_destroy(void *ptr, int flag) { struct disk *dp; struct g_geom *gp; struct g_disk_softc *sc; g_topology_assert(); dp = ptr; gp = dp->d_geom; if (gp != NULL) { sc = gp->softc; if (sc != NULL) sc->dp = NULL; dp->d_geom = NULL; g_wither_geom(gp, ENXIO); } g_free(dp); } /* * We only allow printable characters in disk ident, * the rest is converted to 'x'. */ static void g_disk_ident_adjust(char *ident, size_t size) { char *p, tmp[4], newid[DISK_IDENT_SIZE]; newid[0] = '\0'; for (p = ident; *p != '\0'; p++) { if (isprint(*p)) { tmp[0] = *p; tmp[1] = '\0'; } else { snprintf(tmp, sizeof(tmp), "x%02hhx", *(unsigned char *)p); } if (strlcat(newid, tmp, sizeof(newid)) >= sizeof(newid)) break; } bzero(ident, size); strlcpy(ident, newid, size); } struct disk * disk_alloc(void) { return (g_malloc(sizeof(struct disk), M_WAITOK | M_ZERO)); } void disk_create(struct disk *dp, int version) { if (version != DISK_VERSION) { printf("WARNING: Attempt to add disk %s%d %s", dp->d_name, dp->d_unit, " using incompatible ABI version of disk(9)\n"); printf("WARNING: Ignoring disk %s%d\n", dp->d_name, dp->d_unit); return; } if (version < DISK_VERSION_04) dp->d_flags |= DISKFLAG_LACKS_ROTRATE; KASSERT(dp->d_strategy != NULL, ("disk_create need d_strategy")); KASSERT(dp->d_name != NULL, ("disk_create need d_name")); KASSERT(*dp->d_name != 0, ("disk_create need d_name")); KASSERT(strlen(dp->d_name) < SPECNAMELEN - 4, ("disk name too long")); if (dp->d_devstat == NULL) dp->d_devstat = devstat_new_entry(dp->d_name, dp->d_unit, dp->d_sectorsize, DEVSTAT_ALL_SUPPORTED, DEVSTAT_TYPE_DIRECT, DEVSTAT_PRIORITY_MAX); dp->d_geom = NULL; g_disk_ident_adjust(dp->d_ident, sizeof(dp->d_ident)); g_post_event(g_disk_create, dp, M_WAITOK, dp, NULL); } void disk_destroy(struct disk *dp) { g_cancel_event(dp); dp->d_destroyed = 1; if (dp->d_devstat != NULL) devstat_remove_entry(dp->d_devstat); g_post_event(g_disk_destroy, dp, M_WAITOK, NULL); } void disk_gone(struct disk *dp) { struct g_geom *gp; struct g_provider *pp; gp = dp->d_geom; if (gp != NULL) { pp = LIST_FIRST(&gp->provider); if (pp != NULL) { KASSERT(LIST_NEXT(pp, provider) == NULL, ("geom %p has more than one provider", gp)); g_wither_provider(pp, ENXIO); } } } void disk_attr_changed(struct disk *dp, const char *attr, int flag) { struct g_geom *gp; struct g_provider *pp; + char devnamebuf[128]; gp = dp->d_geom; if (gp != NULL) LIST_FOREACH(pp, &gp->provider, provider) (void)g_attr_changed(pp, attr, flag); + snprintf(devnamebuf, sizeof(devnamebuf), "devname=%s%d", dp->d_name, + dp->d_unit); + devctl_notify("GEOM", "disk", attr, devnamebuf); } void disk_media_changed(struct disk *dp, int flag) { struct g_geom *gp; struct g_provider *pp; gp = dp->d_geom; if (gp != NULL) { pp = LIST_FIRST(&gp->provider); if (pp != NULL) { KASSERT(LIST_NEXT(pp, provider) == NULL, ("geom %p has more than one provider", gp)); g_media_changed(pp, flag); } } } void disk_media_gone(struct disk *dp, int flag) { struct g_geom *gp; struct g_provider *pp; gp = dp->d_geom; if (gp != NULL) { pp = LIST_FIRST(&gp->provider); if (pp != NULL) { KASSERT(LIST_NEXT(pp, provider) == NULL, ("geom %p has more than one provider", gp)); g_media_gone(pp, flag); } } } int disk_resize(struct disk *dp, int flag) { if (dp->d_destroyed || dp->d_geom == NULL) return (0); return (g_post_event(g_disk_resize, dp, flag, NULL)); } static void g_kern_disks(void *p, int flag __unused) { struct sbuf *sb; struct g_geom *gp; char *sp; sb = p; sp = ""; g_topology_assert(); LIST_FOREACH(gp, &g_disk_class.geom, geom) { sbuf_printf(sb, "%s%s", sp, gp->name); sp = " "; } sbuf_finish(sb); } static int sysctl_disks(SYSCTL_HANDLER_ARGS) { int error; struct sbuf *sb; sb = sbuf_new_auto(); g_waitfor_event(g_kern_disks, sb, M_WAITOK, NULL); error = SYSCTL_OUT(req, sbuf_data(sb), sbuf_len(sb) + 1); sbuf_delete(sb); return error; } SYSCTL_PROC(_kern, OID_AUTO, disks, CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0, sysctl_disks, "A", "names of available disks"); Index: stable/10 =================================================================== --- stable/10 (revision 300213) +++ stable/10 (revision 300214) Property changes on: stable/10 ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /head:r298420,298439,298644