[PATCH v4 03/19] nvme: Added a newsysfs attribute appid_store
Hannes Reinecke
hare at suse.de
Mon Nov 16 02:39:41 EST 2020
On 11/9/20 5:23 AM, Muneendra wrote:
> Added a new sysfs attribute appid_store under
> /sys/class/fc/fc_udev_device/*
>
> With this new interface the user can set the application identfier
> in the blkcg associted with cgroup id.
>
> Once the application identifer has set with this interface it allows
> identification of traffic sources at an individual cgroup based
> Applications (ex:virtual machine (VM))level in both host and
> fabric infrastructure(FC).
>
> Below is the interface provided to set the app_id
>
> echo "<cgroupid>:<appid>" >> /sys/class/fc/fc_udev_device/appid_store
> echo "457E:100000109b521d27" >> /sys/class/fc/fc_udev_device/appid_store
>
> Signed-off-by: Muneendra <muneendra.kumar at broadcom.com>
>
> ---
> v4:
> No change
>
> v3:
> Replaced blkcg_set_app_identifier function with blkcg_set_fc_appid
>
> v2:
> New Patch
> ---
> drivers/nvme/host/fc.c | 73 +++++++++++++++++++++++++++++++++++++++++-
> 1 file changed, 72 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/nvme/host/fc.c b/drivers/nvme/host/fc.c
> index eae43bb444e0..6d6cc06fd54a 100644
> --- a/drivers/nvme/host/fc.c
> +++ b/drivers/nvme/host/fc.c
> @@ -9,7 +9,7 @@
> #include <uapi/scsi/fc/fc_els.h>
> #include <linux/delay.h>
> #include <linux/overflow.h>
> -
> +#include <linux/blk-cgroup.h>
> #include "nvme.h"
> #include "fabrics.h"
> #include <linux/nvme-fc-driver.h>
> @@ -3768,10 +3768,81 @@ static ssize_t nvme_fc_nvme_discovery_store(struct device *dev,
>
> return count;
> }
> +
> +/*parse the Cgroup id from a buf and returns the length of cgrpid*/
> +static int fc_parse_cgrpid(const char *buf, u64 *id)
> +{
> + char cgrp_id[16+1];
> + int cgrpid_len, j;
> +
> + memset(cgrp_id, 0x0, sizeof(cgrp_id));
> + for (cgrpid_len = 0, j = 0; cgrpid_len < 17; cgrpid_len++) {
> + if (buf[cgrpid_len] != ':')
> + cgrp_id[cgrpid_len] = buf[cgrpid_len];
> + else {
> + j = 1;
> + break;
> + }
> + }
> + if (!j)
> + return -EINVAL;
> + if (kstrtou64(cgrp_id, 16, id) < 0)
> + return -EINVAL;
> + return cgrpid_len;
> +}
> +
> +/*
> + * fc_update_appid :parses and updates the appid in the blkcg associated with
> + * cgroupid.
> + * @buf: buf contains both cgrpid and appid info
> + * @count: size of the buffer
> + */
> +static int fc_update_appid(const char *buf, size_t count)
> +{
> + u64 cgrp_id;
> + int appid_len = 0;
> + int cgrpid_len = 0;
> + char app_id[APPID_LEN];
> + int ret = 0;
> +
> + if (buf[count-1] == '\n')
> + count--;
> +
> + if ((count > (16+1+APPID_LEN)) || (!strchr(buf, ':')))
> + return -EINVAL;
> +
> + cgrpid_len = fc_parse_cgrpid(buf, &cgrp_id);
> + if (cgrpid_len < 0)
> + return -EINVAL;
> + /*appid len is count - cgrpid_len -1 (: + \n) */
> + appid_len = count - cgrpid_len - 1;
> + if (appid_len > APPID_LEN)
> + return -EINVAL;
> +
> + memset(app_id, 0x0, sizeof(app_id));
> + memcpy(app_id, &buf[cgrpid_len+1], appid_len);
> + ret = blkcg_set_app_identifier(app_id, cgrp_id, sizeof(app_id));
> + if (ret < 0)
> + return ret;
> + return count;
> +}
> +
Right. So you _do_ allow for an arbitrary length for the app_id.
Which means the previous patch had a bug, and you need to allocate
APPID_LEN + 1 for the 'app_id' entry.
> +static ssize_t fc_appid_store(struct device *dev,
> + struct device_attribute *attr, const char *buf, size_t count)
> +{
> + int ret = 0;
> +
> + ret = fc_update_appid(buf, count);
> + if (ret < 0)
> + return -EINVAL;
> + return count;
> +}
> static DEVICE_ATTR(nvme_discovery, 0200, NULL, nvme_fc_nvme_discovery_store);
> +static DEVICE_ATTR(appid_store, 0200, NULL, fc_appid_store);
>
> static struct attribute *nvme_fc_attrs[] = {
> &dev_attr_nvme_discovery.attr,
> + &dev_attr_appid_store.attr,
> NULL
> };
>
>
Reviewed-by: Hannes Reinecke <hare at suse.de>
Cheers,
Hannes
--
Dr. Hannes Reinecke Kernel Storage Architect
hare at suse.de +49 911 74053 688
SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), Geschäftsführer: Felix Imendörffer
More information about the Linux-nvme
mailing list