From: Arthur Simchaev <[email protected]>
The new qTimestamp attribute was added to UFS 4.0 spec, in order to
synchronize timestamp between device logs and the host.The spec recommend
to send this attribute upon device power-on Reset/HW reset or when
switching to Active state (using SSU command). Due to this attribute,
the attribute's max value was extended to 8 bytes. As a result,
the new definition of struct utp_upiu_query_v4_0 was added.
Signed-off-by: Arthur Simchaev <[email protected]>
-----------------
Changes to v2:
- Adressed Bart's comments
- Add missed response variable to ufshcd_set_timestamp_attr
---
drivers/ufs/core/ufshcd.c | 38 ++++++++++++++++++++++++++++++++
include/uapi/scsi/scsi_bsg_ufs.h | 25 +++++++++++++++++++++
include/ufs/ufs.h | 1 +
3 files changed, 64 insertions(+)
diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
index 358b3240b6c5..e2812911e462 100644
--- a/drivers/ufs/core/ufshcd.c
+++ b/drivers/ufs/core/ufshcd.c
@@ -8520,6 +8520,41 @@ static int ufshcd_device_params_init(struct ufs_hba *hba)
return ret;
}
+static void ufshcd_set_timestamp_attr(struct ufs_hba *hba)
+{
+ int err;
+ struct ufs_query_req *request = NULL;
+ struct ufs_query_res *response = NULL;
+ struct ufs_dev_info *dev_info = &hba->dev_info;
+ struct utp_upiu_query_v4_0 *upiu_data;
+
+ if (dev_info->wspecversion < 0x400)
+ return;
+
+ ufshcd_hold(hba);
+
+ mutex_lock(&hba->dev_cmd.lock);
+
+ ufshcd_init_query(hba, &request, &response,
+ UPIU_QUERY_OPCODE_WRITE_ATTR,
+ QUERY_ATTR_IDN_TIMESTAMP, 0, 0);
+
+ request->query_func = UPIU_QUERY_FUNC_STANDARD_WRITE_REQUEST;
+
+ upiu_data = (struct utp_upiu_query_v4_0 *)&request->upiu_req;
+
+ put_unaligned_be64(ktime_get_real_ns(), &upiu_data->osf3);
+
+ err = ufshcd_exec_dev_cmd(hba, DEV_CMD_TYPE_QUERY, QUERY_REQ_TIMEOUT);
+
+ if (err)
+ dev_err(hba->dev, "%s: failed to set timestamp %d\n",
+ __func__, err);
+
+ mutex_unlock(&hba->dev_cmd.lock);
+ ufshcd_release(hba);
+}
+
/**
* ufshcd_add_lus - probe and add UFS logical units
* @hba: per-adapter instance
@@ -8708,6 +8743,8 @@ static int ufshcd_device_init(struct ufs_hba *hba, bool init_dev_params)
ufshcd_set_ufs_dev_active(hba);
ufshcd_force_reset_auto_bkops(hba);
+ ufshcd_set_timestamp_attr(hba);
+
/* Gear up to HS gear if supported */
if (hba->max_pwr_info.is_valid) {
/*
@@ -9741,6 +9778,7 @@ static int __ufshcd_wl_resume(struct ufs_hba *hba, enum ufs_pm_op pm_op)
ret = ufshcd_set_dev_pwr_mode(hba, UFS_ACTIVE_PWR_MODE);
if (ret)
goto set_old_link_state;
+ ufshcd_set_timestamp_attr(hba);
}
if (ufshcd_keep_autobkops_enabled_except_suspend(hba))
diff --git a/include/uapi/scsi/scsi_bsg_ufs.h b/include/uapi/scsi/scsi_bsg_ufs.h
index 2801b65299aa..fd3f9e5ee241 100644
--- a/include/uapi/scsi/scsi_bsg_ufs.h
+++ b/include/uapi/scsi/scsi_bsg_ufs.h
@@ -70,6 +70,31 @@ struct utp_upiu_query {
__be32 reserved[2];
};
+/**
+ * struct utp_upiu_query_v4_0 - upiu request buffer structure for
+ * query request >= UFS 4.0 spec.
+ * @opcode: command to perform B-0
+ * @idn: a value that indicates the particular type of data B-1
+ * @index: Index to further identify data B-2
+ * @selector: Index to further identify data B-3
+ * @osf4: spec field B-5
+ * @osf5: spec field B 6,7
+ * @osf6: spec field DW 8,9
+ * @osf7: spec field DW 10,11
+ */
+struct utp_upiu_query_v4_0 {
+ __u8 opcode;
+ __u8 idn;
+ __u8 index;
+ __u8 selector;
+ __u8 osf3;
+ __u8 osf4;
+ __be16 osf5;
+ __be32 osf6;
+ __be32 osf7;
+ __be32 reserved;
+};
+
/**
* struct utp_upiu_cmd - Command UPIU structure
* @data_transfer_len: Data Transfer Length DW-3
diff --git a/include/ufs/ufs.h b/include/ufs/ufs.h
index 4e8d6240e589..198cb391f9db 100644
--- a/include/ufs/ufs.h
+++ b/include/ufs/ufs.h
@@ -170,6 +170,7 @@ enum attr_idn {
QUERY_ATTR_IDN_WB_BUFF_LIFE_TIME_EST = 0x1E,
QUERY_ATTR_IDN_CURR_WB_BUFF_SIZE = 0x1F,
QUERY_ATTR_IDN_EXT_IID_EN = 0x2A,
+ QUERY_ATTR_IDN_TIMESTAMP = 0x30
};
/* Descriptor idn for Query requests */
--
2.34.1
On Mon, 2023-06-26 at 13:33 +0300, Arthur Simchaev wrote:
>
> +/**
> + * struct utp_upiu_query_v4_0 - upiu request buffer structure for
> + * query request >= UFS 4.0 spec.
> + * @opcode: command to perform B-0
> + * @idn: a value that indicates the particular type of data B-1
> + * @index: Index to further identify data B-2
> + * @selector: Index to further identify data B-3
> + * @osf4: spec field B-5
> + * @osf5: spec field B 6,7
> + * @osf6: spec field DW 8,9
> + * @osf7: spec field DW 10,11
> + */
> +struct utp_upiu_query_v4_0 {
> + __u8 opcode;
> + __u8 idn;
> + __u8 index;
> + __u8 selector;
> + __u8 osf3;
> + __u8 osf4;
> + __be16 osf5;
> + __be32 osf6;
> + __be32 osf7;
> + __be32 reserved;
> +};
Hi Arthur,
In fact, this structure should be a "Transaction Specific Fields" in
Query UPIU for all query functions. By comparing 3.1 and 4.0, you can
use current utp_upiu_query{}, because you only want to pass 64bit TS.
Or you can change utp_upiu_query() to make it more uniform to meet
various query functions.
Kind regards,
Bean
>
> Hi Arthur,
> In fact, this structure should be a "Transaction Specific Fields" in
> Query UPIU for all query functions. By comparing 3.1 and 4.0, you can
> use current utp_upiu_query{}, because you only want to pass 64bit TS.
> Or you can change utp_upiu_query() to make it more uniform to meet
> various query functions.
>
Hi Bean
Actually, in the initial set of patches I changed utp_upiu_query , since the struct defined incorrectly
and I want to change it according to the spec, but according to Bart's comment,
the struct cannot be changed since the definition is in uapi .
See the initial patch sets in the link below:
https://www.spinics.net/lists/linux-scsi/msg181985.html
https://www.spinics.net/lists/linux-scsi/msg182178.html
https://www.spinics.net/lists/linux-scsi/msg181987.html
Please let me know what you think.
Regards
Arthur
On 6/26/23 03:33, Arthur Simchaev wrote:
> +/**
> + * struct utp_upiu_query_v4_0 - upiu request buffer structure for
> + * query request >= UFS 4.0 spec.
> + * @opcode: command to perform B-0
> + * @idn: a value that indicates the particular type of data B-1
> + * @index: Index to further identify data B-2
> + * @selector: Index to further identify data B-3
> + * @osf4: spec field B-5
> + * @osf5: spec field B 6,7
> + * @osf6: spec field DW 8,9
> + * @osf7: spec field DW 10,11
> + */
> +struct utp_upiu_query_v4_0 {
> + __u8 opcode;
> + __u8 idn;
> + __u8 index;
> + __u8 selector;
> + __u8 osf3;
> + __u8 osf4;
> + __be16 osf5;
> + __be32 osf6;
> + __be32 osf7;
> + __be32 reserved;
> +};
Is this structure useful for user space software? If not, please move it
into another header file.
Thanks,
Bart.
> Is this structure useful for user space software? If not, please move it
> into another header file.
Hi Bart
The struct is useful for user space software.
For example, the specific timestamp can be sent by ufs-utils utility to the device.
Regards
Arthur
Hi Bart
Is the patch, OK?
Or do you have additional comments?
Regards
Arthur
> -----Original Message-----
> From: Arthur Simchaev <[email protected]>
> Sent: Thursday, June 29, 2023 4:59 PM
> To: Bart Van Assche <[email protected]>; [email protected]
> Cc: Avri Altman <[email protected]>; Avi Shchislowski
> <[email protected]>; [email protected]; linux-
> [email protected]; [email protected]
> Subject: RE: [RESEND v2] ufs: core: Add support for qTimestamp attribute
>
> CAUTION: This email originated from outside of Western Digital. Do not click
> on links or open attachments unless you recognize the sender and know that
> the content is safe.
>
>
> > Is this structure useful for user space software? If not, please move it
> > into another header file.
> Hi Bart
>
> The struct is useful for user space software.
> For example, the specific timestamp can be sent by ufs-utils utility to the
> device.
>
> Regards
> Arthur
On 6/26/23 03:33, Arthur Simchaev wrote:
> The new qTimestamp attribute was added to UFS 4.0 spec, in order to
> synchronize timestamp between device logs and the host.The spec recommend
> to send this attribute upon device power-on Reset/HW reset or when
> switching to Active state (using SSU command). Due to this attribute,
> the attribute's max value was extended to 8 bytes. As a result,
> the new definition of struct utp_upiu_query_v4_0 was added.
Reviewed-by: Bart Van Assche <[email protected]>
Bart,
>> The new qTimestamp attribute was added to UFS 4.0 spec, in order to
>> synchronize timestamp between device logs and the host.The spec recommend
>> to send this attribute upon device power-on Reset/HW reset or when
>> switching to Active state (using SSU command). Due to this attribute,
>> the attribute's max value was extended to 8 bytes. As a result,
>> the new definition of struct utp_upiu_query_v4_0 was added.
>
> Reviewed-by: Bart Van Assche <[email protected]>
Applied to 6.5/scsi-staging, thanks!
--
Martin K. Petersen Oracle Linux Engineering
On Mon, 26 Jun 2023 13:33:19 +0300, Arthur Simchaev wrote:
> The new qTimestamp attribute was added to UFS 4.0 spec, in order to
> synchronize timestamp between device logs and the host.The spec recommend
> to send this attribute upon device power-on Reset/HW reset or when
> switching to Active state (using SSU command). Due to this attribute,
> the attribute's max value was extended to 8 bytes. As a result,
> the new definition of struct utp_upiu_query_v4_0 was added.
>
> [...]
Applied to 6.5/scsi-fixes, thanks!
[1/1] ufs: core: Add support for qTimestamp attribute
https://git.kernel.org/mkp/scsi/c/24befa92ed47
--
Martin K. Petersen Oracle Linux Engineering