Rework the struct scsi_device inquiry information
Subsystems (3)
cover Rework the struct scsi_device inquiry information
Hi Martin,
The vendor, model and rev members in struct scsi_device are fixed-length strings that are not NUL-terminated. This patch converts these members into NUL-terminated character arrays. This makes it less error-prone to deal with these structure members. The patches in this series have been implemented such that the number of lines changed and the risk for regressions is minimized.
Please consider this patch series for the next merge window.
Thanks,
Bart.
Bart Van Assche (2):
scsi: core, target: Move three constants into <scsi/scsi_common.h>
scsi: core: Convert inquiry information
drivers/hwmon/drivetemp.c | 5 +---- drivers/scsi/scsi_scan.c | 12 ++++++------ include/scsi/scsi_common.h | 4 ++++ include/scsi/scsi_device.h | 7 ++++--- include/target/target_core_base.h | 5 +---- 5 files changed, 16 insertions(+), 17 deletions(-)
PATCH 1/2 scsi: core, target: Move three constants into <scsi/scsi_common.h>
Prepare for using these constants in the SCSI core.
--- include/scsi/scsi_common.h | 4 ++++ include/target/target_core_base.h | 5 +---- 2 files changed, 5 insertions(+), 4 deletions(-)
include/scsi/scsi_common.h +4 -0
@@ -10,6 +10,10 @@#include <uapi/linux/pr.h> #include <scsi/scsi_proto.h> +#define INQUIRY_VENDOR_LEN 8 +#define INQUIRY_MODEL_LEN 16 +#define INQUIRY_REVISION_LEN 4 +enum scsi_pr_type {SCSI_PR_WRITE_EXCLUSIVE =0x01 ,SCSI_PR_EXCLUSIVE_ACCESS =0x03 ,
include/target/target_core_base.h +1 -4
@@ -8,6 +8,7 @@#include <linux/percpu-refcount.h> #include <linux/semaphore.h> /* struct semaphore */ #include <linux/completion.h> +#include <scsi/scsi_common.h> #define TARGET_CORE_VERSION "v5.0"
@@ -46,10 +47,6 @@/* Used by transport_get_inquiry_vpd_device_ident() */ #define INQUIRY_VPD_DEVICE_IDENTIFIER_LEN 254 -#define INQUIRY_VENDOR_LEN 8 -#define INQUIRY_MODEL_LEN 16 -#define INQUIRY_REVISION_LEN 4 -/* Attempts before moving from SHORT to LONG */ #define PYX_TRANSPORT_WINDOW_CLOSED_THRESHOLD 3#define PYX_TRANSPORT_WINDOW_CLOSED_WAIT_SHORT 3/* In milliseconds */
Re: [PATCH 1/2] scsi: core, target: Move three constants into <scsi/scsi_common.h>
Quoted 3 lines · Bart Van Assche
Prepare for using these constants in the SCSI core.
Signed-off-by Bart Van Assche <[email protected]>
Looks OK to me.
Signature · 2 lines
Damien Le Moal Western Digital Research
Re: [PATCH 1/2] scsi: core, target: Move three constants into <scsi/scsi_common.h>
In reply to Bart Van AsscheQuoted 8 lines · Bart Van Assche
Prepare for using these constants in the SCSI core.
Signed-off-by Bart Van Assche <[email protected]>--- include/scsi/scsi_common.h | 4 ++++ include/target/target_core_base.h | 5 +---- 2 files changed, 5 insertions(+), 4 deletions(-)
Cheers,
Hannes
Signature · 4 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
PATCH 2/2 scsi: core: Convert inquiry information
In reply to Bart Van AsscheCurrently the vendor, model, and revision members of struct scsi_device are pointers to fixed-length strings that are not NUL-terminated. Fixed-precision format specifiers (e.g., "%.8s") are required whenever they are printed and strncmp() must be used to compare these fields. This is error-prone.
Convert these fields to fixed-size character arrays within struct scsi_device. Remove an !sdev->model check because sdev->model is now guaranteed not to be NULL.
This patch fixes a bug in the qla2xxx driver. It makes the following code safe:
if (state_flags & BIT_4)
scmd_printk(KERN_WARNING, cp,
"Unsupported device '%s' found.\n", cp->device->vendor);
--- drivers/hwmon/drivetemp.c | 5 +---- drivers/scsi/scsi_scan.c | 12 ++++++------ include/scsi/scsi_device.h | 7 ++++--- 3 files changed, 11 insertions(+), 13 deletions(-)
drivers/hwmon/drivetemp.c +1 -4
@@ -306,13 +306,10 @@ static bool drivetemp_sct_avoid(struct drivetemp_data *st)struct scsi_device *sdev = st->sdev;unsigned int ctr; -if (!sdev->model) -return false; -/* * The "model" field contains just the raw SCSI INQUIRY response * "product identification" field, which has a width of 16 bytes. -* This field is space-filled, but is NOT NULL-terminated. +* This field is space-filled and NUL-terminated. */ for (ctr =0 ; ctr <ARRAY_SIZE (sct_avoid_models); ctr++) if (!strncmp(sdev->model, sct_avoid_models[ctr],
drivers/scsi/scsi_scan.c +6 -6
@@ -292,9 +292,9 @@ static struct scsi_device *scsi_alloc_sdev(struct scsi_target *starget,if (!sdev) goto out; - sdev->vendor = scsi_null_device_strs; - sdev->model = scsi_null_device_strs; - sdev->rev = scsi_null_device_strs; +strscpy (sdev->vendor, scsi_null_device_strs); +strscpy (sdev->model, scsi_null_device_strs); +strscpy (sdev->rev, scsi_null_device_strs); sdev->host = shost; sdev->queue_ramp_up_period =SCSI_DEFAULT_RAMP_UP_PERIOD ; sdev->id = starget->id;
@@ -905,9 +905,9 @@ static int scsi_add_lun(struct scsi_device *sdev, unsigned char *inq_result,if (sdev->inquiry ==NULL )return SCSI_SCAN_NO_RESPONSE ; - sdev->vendor = (char *) (sdev->inquiry +8 ); - sdev->model = (char *) (sdev->inquiry +16 ); - sdev->rev = (char *) (sdev->inquiry +32 ); +strscpy (sdev->vendor, sdev->inquiry +8 ); +strscpy (sdev->model, sdev->inquiry +16 ); +strscpy (sdev->rev, sdev->inquiry +32 ); sdev->is_ata =strncmp (sdev->vendor,"ATA " ,8 ) ==0 ;if (sdev->is_ata) {
include/scsi/scsi_device.h +4 -3
@@ -7,6 +7,7 @@#include <linux/workqueue.h> #include <linux/blk-mq.h> #include <scsi/scsi.h> +#include <scsi/scsi_common.h> #include <linux/atomic.h> #include <linux/sbitmap.h>
@@ -137,9 +138,9 @@ struct scsi_device {struct mutex inquiry_mutex;unsigned char inquiry_len;/* valid bytes in 'inquiry' */ unsigned char * inquiry;/* INQUIRY response data */ -const char * vendor;/* [back_compat] point into 'inquiry' ... */ -const char * model;/* ... after scan; point to static string */ -const char * rev;/* ... "nullnullnullnull" before scan */ +char vendor[INQUIRY_VENDOR_LEN +1 ]; +char model[INQUIRY_MODEL_LEN +1 ]; +char rev[INQUIRY_REVISION_LEN +1 ];#define SCSI_DEFAULT_VPD_LEN 255/* default SCSI VPD page size (max) */ struct scsi_vpd __rcu *vpd_pg0;
Re: [PATCH 2/2] scsi: core: Convert inquiry information
Quoted 2 lines · Bart Van Assche
Currently the vendor, model, and revision members of struct
scsi_deviceare pointers to fixed-length strings that are not NUL-terminated.
s/NUL/NULL
Quoted 29 lines
Fixed-precision format specifiers (e.g., "%.8s") are required whenever they are printed and
strncmp()must be used to compare these fields. This is error-prone.Convert these fields to fixed-size character arrays within struct
scsi_device. Remove an !sdev->model check becausesdev->modelis now guaranteed not to be NULL.This patch fixes a bug in the qla2xxx driver. It makes the following code safe:
if (
state_flags&BIT_4)
scmd_printk(KERN_WARNING, cp,
"Unsupported device '%s' found.\n",cp->device->vendor);Signed-off-by Bart Van Assche <[email protected]>--- drivers/hwmon/drivetemp.c | 5 +---- drivers/scsi/scsi_scan.c | 12 ++++++------ include/scsi/scsi_device.h | 7 ++++--- 3 files changed, 11 insertions(+), 13 deletions(-)drivers/hwmon/drivetemp.c +0
-0@@ -306,13 +306,10 @@ static bool drivetemp_sct_avoid(struct drivetemp_data *st)struct scsi_device *sdev = st->sdev;
drivers/hwmon/drivetemp.c static bool drivetemp_sct_avoid(struct drivetemp_data *st) · line 312
Quoted 10 lines
drivers/hwmon/drivetemp.c +1
-4unsigned int ctr; -if (!sdev->model) -return false; -/* * The "model" field contains just the raw SCSI INQUIRY response * "product identification" field, which has a width of 16 bytes. -* This field is space-filled, but is NOT NULL-terminated. +* This field is space-filled and NUL-terminated.
s/NUL/NULL
Quoted 58 lines
drivers/hwmon/drivetemp.c +0
-0*/ for (ctr =0 ; ctr <ARRAY_SIZE (sct_avoid_models); ctr++) if (!strncmp(sdev->model, sct_avoid_models[ctr],drivers/scsi/scsi_scan.c +6
-6@@ -292,9 +292,9 @@ static struct scsi_device *scsi_alloc_sdev(struct scsi_target *starget,if (!sdev) goto out; - sdev->vendor = scsi_null_device_strs; - sdev->model = scsi_null_device_strs; - sdev->rev = scsi_null_device_strs; +strscpy (sdev->vendor, scsi_null_device_strs); +strscpy (sdev->model, scsi_null_device_strs); +strscpy (sdev->rev, scsi_null_device_strs); sdev->host = shost; sdev->queue_ramp_up_period =SCSI_DEFAULT_RAMP_UP_PERIOD ; sdev->id = starget->id;@@ -905,9 +905,9 @@ static int scsi_add_lun(struct scsi_device *sdev, unsigned char *inq_result,if (sdev->inquiry ==NULL )return SCSI_SCAN_NO_RESPONSE ; - sdev->vendor = (char *) (sdev->inquiry +8 ); - sdev->model = (char *) (sdev->inquiry +16 ); - sdev->rev = (char *) (sdev->inquiry +32 ); +strscpy (sdev->vendor, sdev->inquiry +8 ); +strscpy (sdev->model, sdev->inquiry +16 ); +strscpy (sdev->rev, sdev->inquiry +32 ); sdev->is_ata =strncmp (sdev->vendor,"ATA " ,8 ) ==0 ;if (sdev->is_ata) {include/scsi/scsi_device.h +4
-3@@ -7,6 +7,7 @@#include <linux/workqueue.h> #include <linux/blk-mq.h> #include <scsi/scsi.h> +#include <scsi/scsi_common.h> #include <linux/atomic.h> #include <linux/sbitmap.h> @@ -137,9 +138,9 @@ struct scsi_device {struct mutex inquiry_mutex;unsigned char inquiry_len;/* valid bytes in 'inquiry' */ unsigned char * inquiry;/* INQUIRY response data */ -const char * vendor;/* [back_compat] point into 'inquiry' ... */ -const char * model;/* ... after scan; point to static string */ -const char * rev;/* ... "nullnullnullnull" before scan */ +char vendor[INQUIRY_VENDOR_LEN +1 ]; +char model[INQUIRY_MODEL_LEN +1 ]; +char rev[INQUIRY_REVISION_LEN +1 ];#define SCSI_DEFAULT_VPD_LEN 255/* default SCSI VPD page size (max) */ struct scsi_vpd __rcu *vpd_pg0;
Signature · 2 lines
Damien Le Moal Western Digital Research
Re: [PATCH 2/2] scsi: core: Convert inquiry information
Quoted 5 lines · Damien Le Moal
Quoted 2 lines · Bart Van Assche
Currently the vendor, model, and revision members of struct
scsi_deviceare pointers to fixed-length strings that are not NUL-terminated.s/NUL/NULL
Really? NUL is the correct spelling according to
https://en.wikipedia.org/wiki/ASCII and also according to any other ASCII table I have ever seen.
Thanks,
Bart.
Guenter Roeck
maintainerRe: [PATCH 2/2] scsi: core: Convert inquiry information
Quoted 10 lines · Bart Van Assche
Quoted 5 lines · Damien Le Moal
Quoted 2 lines · Bart Van Assche
Currently the vendor, model, and revision members of struct
scsi_deviceare pointers to fixed-length strings that are not NUL-terminated.s/NUL/NULL
Really? NUL is the correct spelling according to
https://en.wikipedia.org/wiki/ASCII and also according to any other ASCII table I have ever seen.
Maybe the confusion is NULL pointer (or NULL-terminated array) vs. NUL-terminated string ?
Guenter
Re: [PATCH 2/2] scsi: core: Convert inquiry information
Quoted 6 lines · Guenter Roeck
Quoted 5 lines · Bart Van Assche
Quoted 4 lines · Damien Le Moal
Quoted 2 lines · Bart Van Assche
Currently the vendor, model, and revision members of struct
scsi_deviceare pointers to fixed-length strings that are not NUL-terminated.
Quoted 8 lines
Quoted 5 lines
Quoted 1 line
s/NUL/NULL
Really? NUL is the correct spelling according to
https://en.wikipedia.org/wiki/ASCII and also according to any other ASCII table I have ever seen.Maybe the confusion is NULL pointer (or NULL-terminated array) vs. NUL-terminated string ?
Ah, yes, indeed. Please ignore my comment.
Signature · 2 lines
Damien Le Moal Western Digital Research
Re: [PATCH 2/2] scsi: core: Convert inquiry information
In reply to Bart Van AsscheQuoted 64 lines · Bart Van Assche
Currently the vendor, model, and revision members of struct
scsi_deviceare pointers to fixed-length strings that are not NUL-terminated. Fixed-precision format specifiers (e.g., "%.8s") are required whenever they are printed andstrncmp()must be used to compare these fields. This is error-prone.Convert these fields to fixed-size character arrays within struct
scsi_device. Remove an !sdev->model check becausesdev->modelis now guaranteed not to be NULL.This patch fixes a bug in the qla2xxx driver. It makes the following code safe:
if (
state_flags&BIT_4)
scmd_printk(KERN_WARNING, cp,
"Unsupported device '%s' found.\n",cp->device->vendor);Signed-off-by Bart Van Assche <[email protected]>--- drivers/hwmon/drivetemp.c | 5 +---- drivers/scsi/scsi_scan.c | 12 ++++++------ include/scsi/scsi_device.h | 7 ++++--- 3 files changed, 11 insertions(+), 13 deletions(-)drivers/hwmon/drivetemp.c +1
-4@@ -306,13 +306,10 @@ static bool drivetemp_sct_avoid(struct drivetemp_data *st)struct scsi_device *sdev = st->sdev;unsigned int ctr; -if (!sdev->model) -return false; -/* * The "model" field contains just the raw SCSI INQUIRY response * "product identification" field, which has a width of 16 bytes. -* This field is space-filled, but is NOT NULL-terminated. +* This field is space-filled and NUL-terminated. */ for (ctr =0 ; ctr <ARRAY_SIZE (sct_avoid_models); ctr++) if (!strncmp(sdev->model, sct_avoid_models[ctr],drivers/scsi/scsi_scan.c +3
-3@@ -292,9 +292,9 @@ static struct scsi_device *scsi_alloc_sdev(struct scsi_target *starget,if (!sdev) goto out; - sdev->vendor = scsi_null_device_strs; - sdev->model = scsi_null_device_strs; - sdev->rev = scsi_null_device_strs; +strscpy (sdev->vendor, scsi_null_device_strs); +strscpy (sdev->model, scsi_null_device_strs); +strscpy (sdev->rev, scsi_null_device_strs); sdev->host = shost; sdev->queue_ramp_up_period =SCSI_DEFAULT_RAMP_UP_PERIOD ; sdev->id = starget->id;@@ -905,9 +905,9 @@ static int scsi_add_lun(struct scsi_device *sdev, unsigned char *inq_result,if (sdev->inquiry ==NULL )return SCSI_SCAN_NO_RESPONSE ;
drivers/scsi/scsi_scan.c static int scsi_add_lun(struct scsi_device *sdev, unsigned char *inq_result, · line 913
Quoted 10 lines
drivers/scsi/scsi_scan.c +3
-3- sdev->vendor = (char *) (sdev->inquiry +8 ); - sdev->model = (char *) (sdev->inquiry +16 ); - sdev->rev = (char *) (sdev->inquiry +32 ); +strscpy (sdev->vendor, sdev->inquiry +8 ); +strscpy (sdev->model, sdev->inquiry +16 ); +strscpy (sdev->rev, sdev->inquiry +32 ); sdev->is_ata =strncmp (sdev->vendor,"ATA " ,8 ) ==0 ;if (sdev->is_ata) {
Question is whether we shouldn't make this generic, ie treat 'inquiry'
as a temporary blob, copy things over to fields in 'sdev', and then
free the 'inquiry' blob again.
There are soo many things tacked onto the standard inquiry data (especially for storage array trying to mimic SCSI-2 inquiry data), that we're better of copying over only fields which we _know_.
_And_ it'll save us a permanent data allocation for the scsi device...
Hmm?
Cheers,
Hannes
Signature · 4 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
Re: [PATCH 2/2] scsi: core: Convert inquiry information
Quoted 8 lines · Hannes Reinecke
Question is whether we shouldn't make this generic, ie treat 'inquiry'
as a temporary blob, copy things over to fields in 'sdev', and then
free the 'inquiry' blob again.
There are soo many things tacked onto the standard inquiry data (especially for storage array trying to mimic SCSI-2 inquiry data), that we're better of copying over only fields which we _know_.
_And_ it'll save us a permanent data allocation for the scsi device...Hmm?
Are you perhaps suggesting to remove the inquiry sysfs attribute? From scsi_sysfs.c:
static ssize_t show_inquiry(struct file *filep, struct kobject *kobj,
const struct bin_attribute *bin_attr, char *buf, loff_t off, size_t count)
{
struct device *dev = kobj_to_dev(kobj);
struct scsi_device *sdev = to_scsi_device(dev);
if (!sdev->inquiry)
return -EINVAL;
return memory_read_from_buffer(buf, count, &off, sdev->inquiry,
sdev->inquiry_len);
}Bart.
Re: [PATCH 2/2] scsi: core: Convert inquiry information
Quoted 20 lines · Bart Van Assche
Quoted 8 lines · Hannes Reinecke
Question is whether we shouldn't make this generic, ie treat 'inquiry'
as a temporary blob, copy things over to fields in 'sdev', and then
free the 'inquiry' blob again.
There are soo many things tacked onto the standard inquiry data (especially for storage array trying to mimic SCSI-2 inquiry data), that we're better of copying over only fields which we _know_.
_And_ it'll save us a permanent data allocation for the scsi device...Hmm?
Are you perhaps suggesting to remove the inquiry sysfs attribute? From
scsi_sysfs.c:static
ssize_tshow_inquiry(struct file *filep, struct kobject *kobj,
const structbin_attribute*bin_attr, char *buf,loff_toff,size_tcount)
{
struct device *dev =kobj_to_dev(kobj);
structscsi_device*sdev =to_scsi_device(dev);
Quoted 7 lines
if (!sdev->inquiry)
return-EINVAL;return
memory_read_from_buffer(buf, count, &off, sdev->inquiry, sdev->inquiry_len);
}
Ah. no. Emphatically not; we need that for the udev rules.
So ignore my comment.
Cheers,
Hannes
Signature · 4 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich
James Bottomley
maintainerRe: [PATCH 2/2] scsi: core: Convert inquiry information
In reply to Bart Van AsscheQuoted 11 lines · Bart Van Assche
drivers/scsi/scsi_scan.c +0
-0@@ -905,9 +905,9 @@ static int scsi_add_lun(struct scsi_device *sdev, unsigned char *inq_result, if (sdev->inquiry == NULL) return SCSI_SCAN_NO_RESPONSE;drivers/scsi/scsi_scan.c +3
-3- sdev->vendor = (char *) (sdev->inquiry +8 ); - sdev->model = (char *) (sdev->inquiry +16 ); - sdev->rev = (char *) (sdev->inquiry +32 ); +strscpy (sdev->vendor, sdev->inquiry +8 ); +strscpy (sdev->model, sdev->inquiry +16 ); +strscpy (sdev->rev, sdev->inquiry +32 );
Quoted 3 lines
sdev->is_ata=strncmp(sdev->vendor, "ATA ", 8)== 0; if (sdev->is_ata) {
If you've gone to all the trouble to lift the lengths out as #defines, shouldn't we do the same with the offsets?
Regards,
James
Re: [PATCH 2/2] scsi: core: Convert inquiry information
Quoted 13 lines · James Bottomley
Quoted 12 lines · Bart Van Assche
drivers/scsi/scsi_scan.c +3
-3@@ -905,9 +905,9 @@ static int scsi_add_lun(struct scsi_device *sdev, unsigned char *inq_result, if (sdev->inquiry == NULL) return SCSI_SCAN_NO_RESPONSE; - sdev->vendor = (char *) (sdev->inquiry +8 ); - sdev->model = (char *) (sdev->inquiry +16 ); - sdev->rev = (char *) (sdev->inquiry +32 ); +strscpy (sdev->vendor, sdev->inquiry +8 ); +strscpy (sdev->model, sdev->inquiry +16 ); +strscpy (sdev->rev, sdev->inquiry +32 );
Quoted 5 lines
Quoted 2 lines
drivers/scsi/scsi_scan.c +0
-0sdev->is_ata = strncmp(sdev->vendor, "ATA ", 8) == 0;if (
sdev->is_ata) {If you've gone to all the trouble to lift the lengths out as #defines, shouldn't we do the same with the offsets?
Hi James,
I will introduce symbolic names for the offsets in the next version of this patch series.
Thanks,
Bart.