[PATCH] [Target_Core_Mod/ConfigFS]: Added initial support for filedescriptors for se_device_t

From: Nicholas A. Bellinger
Date: Tue Oct 21 2008 - 16:25:12 EST

Next message: Nicholas A. Bellinger: "[PATCH] [Target_Core_Mod/IBLOCK]: Added support for->create_virtdevice_from_fd()"
Previous message: Serge E. Hallyn: "Re: [RFC v7][PATCH 2/9] General infrastructure for checkpointrestart"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hi again Folks,

Here is the first commit using Joel's recommendations allow
target_core_mod configfs generic infrastructure to allow subsystem
plugins from target_core_mod to handle FDs passed using TYPE_DISK for
Linux/SCSI and Linux/BLOCK and TYPE_CDROM with Linux/SCSI.

Below, se_subsystem_dev_t->create_virtdevice_from_fd() method into Linux
storage subsystem APIs is added to;

/sys/kernel/config/target/core/$PLUGIN_HBA/$STORAGE_OBJECT/fd

This commit depends upon the changes to target_core_mod/IBLOCK and
target_core_mod/PSCSI:

http://git.kernel.org/?p=linux/kernel/git/nab/lio-core-2.6.git;a=commitdiff;h=9c6293c1bfe82d04a215fab22a3a850ea879a137
http://git.kernel.org/?p=linux/kernel/git/nab/lio-core-2.6.git;a=commitdiff;h=bd4227278bbb6ec911a15b380f2479782cbe94ce

There are a few FIXMEs below.. Also, using target_core_mod/PSCSI with
*NON* struct block_device backed struct scsi_devices (eg: character
devices like TYPE_TAPE, TYPE_MEDIUM_CHANGER, etc) is currently not
implemented in the target_core_mod/PSCSI commit above. If anyone can
tell me how to get struct scsi_device from /dev/st0 it would be much
appericated..

Also, the sr0 device appears to need the medium in the drive before it
can great a READ FD from /dev/sr0 initially (I have recalled seeing
something like this in the past, eg: where struct scsi_device is not
attached until the first struct scsi_cmnd come down through Linux/SCSI's
->queuecommand().. It (Linux/SCSI with TYPE_CDROM) has worked like this
as long as I have known, is it possible to get access to TYPE_CDROM's FD
through /dev/sr0 to struct scsi_device before the medium is in the
drive..? Should I just restrict the FD method to TYPE_DISK for
Linux/BLOCK and Linux/SCSI..?

Comments..?

--nab

>From 362d4ef391699b0282dfbef3da633a561d9cbde7 Mon Sep 17 00:00:00 2001
From: Nicholas Bellinger <nab@xxxxxxxxxxxxxxx>
Date: Mon, 20 Oct 2008 23:21:43 -0700
Subject: [PATCH] [Target_Core_Mod/ConfigFS]: Added initial support for file descriptors for se_device_t

Added support for passing in file descriptors for referencing Linux storage objects
in the generic target core. This patch adds the method to the subsystem API
se_subsystem_api_t->create_virtdevice_from_fd() and the configfs entries
$TARGET/$HBA_$ID/$STORAGE_OBJECT/fd.

Currently Linux/SCSI and Linux/BLOCK are supported, here are some examples:

mkdir -p $TARGET/iblock_0/lvm_test0
exec 3<>/dev/lio-test/test0 # RW
echo 3 > $TARGET/iblock_0/lvm_test0/fd
exec 3>&-

<SNIP>

Target_Core_ConfigFS: Located se_plugin: ced17090 plugin_name: iblock hba_type: 4 plugin_dep_id: 0
CORE_HBA[0] - Linux-iSCSI.org iBlock HBA Driver v3.0 on Generic Target Core Stack v3.0.0
CORE_HBA[0] - Attached iBlock HBA: 0 to Generic Target Core TCQ Depth: 512
CORE_HBA[0] - Attached HBA to Generic Target Core
IBLOCK: Allocated ib_dev for lvm_test0
Target_Core_ConfigFS: Allocated se_subsystem_dev_t: cefe6400 se_dev_su_ptr: cf9fd200
IBLOCK: Claiming struct block_device: cf63cf00
IBLOCK: Created bio_set() for major/minor: 254:2
CORE_iBLOCK[0] - Activating Device with TCQ: 0 at Major: 254 Minor 2
Vendor: LIO-ORG Model: IBLOCK Revision: v3.0
Type: Direct-Access ANSI SCSI revision: 02
T10 EVPD Unit Serial Number: 1234567890:0_254_2
T10 EVPD Device Identifier: LIO-ORG IBLOCK:1234567890:0_254_2
Target_Core_ConfigFS: Registered iblock se_dev->se_dev_ptr: cf350000 from fd

<SNIP>

mkdir -p $TARGET/pscsi_0/sdd
exec 3<>/dev/sdd # RW
echo 3 > $TARGET/pscsi_0/sdd/fd
exec 3>&-

<SNIP>

Target_Core_ConfigFS: Located se_plugin: ced17024 plugin_name: pscsi hba_type: 1 plugin_dep_id: 0
CORE_HBA[1] - Linux-iSCSI.org Parallel SCSI HBA Driver v3.0 on Generic Target Core Stack v3.0.0
CORE_HBA[1] - MPT SPI Host
CORE_HBA[1] - Attached Parallel SCSI HBA to Generic Target Core with TCQ Depth: 127 MaxSectors: 8192
CORE_HBA[1] - Attached HBA to Generic Target Core
PSCSI: Allocated pdv: cfb42800 for sdd
Target_Core_ConfigFS: Allocated se_subsystem_dev_t: cf86fe80 se_dev_su_ptr: cfb42800
PSCSI: Claiming cf87f800 Major:Minor - 8:48
CORE_PSCSI[0] - Activating Device with TCQ: 32 at Parallel SCSI Location (Channel/Target/LUN) 0/3/0
Vendor: VMware, Model: VMware Virtual S Revision: 1.0
Type: Direct-Access ANSI SCSI revision: 02
Parallel SCSI Status Byte exception - ITT: 0x00000000 Task: cee6d600 CDB: 0x12 Result: 0x08000002
CORE_PSCSI[0] - Added TYPE_DISK for 0:3:0
Target_Core_ConfigFS: Registered pscsi se_dev->se_dev_ptr: ced6ec00 from fd

<SNIP>

exec 3</dev/sr0 # RO
echo 3 > $TARGET/pscsi_1/sr0/fd
exec 3>&-

<SNIP>

Target_Core_ConfigFS: Located se_plugin: ced17024 plugin_name: pscsi hba_type: 1 plugin_dep_id: 1
CORE_HBA[2] - Linux-iSCSI.org Parallel SCSI HBA Driver v3.0 on Generic Target Core Stack v3.0.0
CORE_HBA[2] - MPT SPI Host
CORE_HBA[2] - Attached Parallel SCSI HBA to Generic Target Core with TCQ Depth: 127 MaxSectors: 8192
CORE_HBA[2] - Attached HBA to Generic Target Core
PSCSI: Allocated pdv: cfb42640 for sr0
Target_Core_ConfigFS: Allocated se_subsystem_dev_t: ced70f00 se_dev_su_ptr: cfb42640
CORE_PSCSI[1] - Activating Device with TCQ: 32 at Parallel SCSI Location (Channel/Target/LUN) 0/0/0
Vendor: VMware, Model: Virtual CD-ROM Revision: 1.0
Type: CD-ROM ANSI SCSI revision: 02
Parallel SCSI Status Byte exception - ITT: 0x00000000 Task: cfa79200 CDB: 0x12 Result: 0x08000002
CORE_PSCSI[1] - Added Type: CD-ROM for 0:0:0
Target_Core_ConfigFS: Registered pscsi se_dev->se_dev_ptr: ceed1800 from fd

<SNIP>

There a few FIXMEs:

*) Determine long term status of struct gendisk->driverfs_dev for
locating struct scsi_device from struct inode for Linux/SCSI.
*) Add support for struct scsi_device's backed by character device.
*) SCSI CDROM can only be attached through the fd method when medium
is in the drive.

Signed-off-by: Nicholas A. Bellinger <nab@xxxxxxxxxxxxxxx>
---
drivers/lio-core/target_core_configfs.c | 54 +++++++++++++++++++++++++++++-
drivers/lio-core/target_core_transport.h | 4 ++
2 files changed, 57 insertions(+), 1 deletions(-)

diff --git a/drivers/lio-core/target_core_configfs.c b/drivers/lio-core/target_core_configfs.c
index 268ab38..264dcdb 100644
--- a/drivers/lio-core/target_core_configfs.c
+++ b/drivers/lio-core/target_core_configfs.c
@@ -574,9 +574,57 @@ static struct target_core_configfs_attribute target_core_attr_dev_control = {
.store = target_core_store_dev_control,
};

+static ssize_t target_core_store_dev_fd (void *p, const char *page, size_t count)
+{
+ se_subsystem_dev_t *se_dev = (se_subsystem_dev_t *)p;
+ se_device_t *dev;
+ se_hba_t *hba = se_dev->se_dev_hba;
+ se_subsystem_api_t *t;
+ int ret = 0;
+
+ if (se_dev->se_dev_ptr) {
+ printk(KERN_ERR "se_dev->se_dev_ptr already set, ignoring fd request\n");
+ return(-EEXIST);
+ }
+
+ t = (se_subsystem_api_t *)plugin_get_obj(PLUGIN_TYPE_TRANSPORT, hba->type, &ret);
+ if (!t || (ret != 0))
+ return(-EINVAL);
+
+ if (!(t->create_virtdevice_from_fd)) {
+ printk(KERN_ERR "se_subsystem_api_t->create_virtdevice_from_fd()"
+ " NULL for: %s\n", hba->transport->name);
+ return(-EINVAL);
+ }
+ /*
+ * The subsystem PLUGIN is responsible for calling target_core_mod
+ * symbols to claim the underlying struct block_device for TYPE_DISK.
+ */
+ dev = t->create_virtdevice_from_fd(se_dev, page);
+ if (!(dev) || IS_ERR(dev))
+ goto out;
+
+ se_dev->se_dev_ptr = dev;
+
+ printk("Target_Core_ConfigFS: Registered %s se_dev->se_dev_ptr: %p"
+ " from fd\n", hba->transport->name, se_dev->se_dev_ptr);
+ return(count);
+out:
+ return(-EINVAL);
+}
+
+static struct target_core_configfs_attribute target_core_attr_dev_fd = {
+ .attr = { .ca_owner = THIS_MODULE,
+ .ca_name = "fd",
+ .ca_mode = S_IWUSR },
+ .show = NULL,
+ .store = target_core_store_dev_fd,
+};
+
static ssize_t target_core_store_dev_enable (void *p, const char *page, size_t count)
{
se_subsystem_dev_t *se_dev = (se_subsystem_dev_t *)p;
+ se_device_t *dev;
se_hba_t *hba = se_dev->se_dev_hba;
se_subsystem_api_t *t;
char *ptr;
@@ -598,10 +646,13 @@ static ssize_t target_core_store_dev_enable (void *p, const char *page, size_t c
if (t->check_configfs_dev_params(hba, se_dev) < 0)
return(-EINVAL);

- if (!(se_dev->se_dev_ptr = t->create_virtdevice(hba, se_dev->se_dev_su_ptr)))
+ dev = t->create_virtdevice(hba, se_dev->se_dev_su_ptr);
+ if (!(dev) || IS_ERR(dev))
return(-EINVAL);

+ se_dev->se_dev_ptr = dev;
printk("Target_Core_ConfigFS: Registered se_dev->se_dev_ptr: %p\n", se_dev->se_dev_ptr);
+
return(count);
}

@@ -616,6 +667,7 @@ static struct target_core_configfs_attribute target_core_attr_dev_enable = {
static struct configfs_attribute *lio_core_dev_attrs[] = {
&target_core_attr_dev_info.attr,
&target_core_attr_dev_control.attr,
+ &target_core_attr_dev_fd.attr,
&target_core_attr_dev_enable.attr,
NULL,
};
diff --git a/drivers/lio-core/target_core_transport.h b/drivers/lio-core/target_core_transport.h
index 7f96176..078750b 100644
--- a/drivers/lio-core/target_core_transport.h
+++ b/drivers/lio-core/target_core_transport.h
@@ -427,6 +427,10 @@ typedef struct se_subsystem_api_s {
*/
ssize_t (*show_configfs_dev_params)(se_hba_t *, se_subsystem_dev_t *, char *);
/*
+ * create_virtdevice_from-fd():
+ */
+ se_device_t *(*create_virtdevice_from_fd)(se_subsystem_dev_t *, const char *);
+ /*
* check_dev_params():
*/
int (*check_dev_params)(se_hba_t *, struct iscsi_target *, se_dev_transport_info_t *);
--
1.5.4.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Next message: Nicholas A. Bellinger: "[PATCH] [Target_Core_Mod/IBLOCK]: Added support for->create_virtdevice_from_fd()"
Previous message: Serge E. Hallyn: "Re: [RFC v7][PATCH 2/9] General infrastructure for checkpointrestart"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]