| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Fri, 17 Feb 2023 14:08:53 +0800
From: Ziqi Chen <quic_ziqichen@...cinc.com>
To: Steven Rostedt <rostedt@...dmis.org>
CC: <quic_asutoshd@...cinc.com>, <quic_cang@...cinc.com>,
<bvanassche@....org>, <mani@...nel.org>,
<stanley.chu@...iatek.com>, <adrian.hunter@...el.com>,
<beanhuo@...ron.com>, <avri.altman@....com>,
<junwoo80.lee@...sung.com>, <martin.petersen@...cle.com>,
<linux-scsi@...r.kernel.org>,
Alim Akhtar <alim.akhtar@...sung.com>,
"James E.J. Bottomley" <jejb@...ux.ibm.com>,
Masami Hiramatsu <mhiramat@...nel.org>,
open list <linux-kernel@...r.kernel.org>,
"open list:TRACING" <linux-trace-kernel@...r.kernel.org>
Subject: Re: [PATCH v1] scsi: ufs: core: Add trace event for MCQ
On 2/17/2023 1:22 AM, Steven Rostedt wrote:
> On Thu, 16 Feb 2023 10:45:55 +0800
> Ziqi Chen <quic_ziqichen@...cinc.com> wrote:
>
>> Added a new trace event to record MCQ relevant information
>> for each request in MCQ mode, include hardware queue ID,
>> SQ tail slot, CQ head slot and CQ tail slot.
>>
>> Signed-off-by: Ziqi Chen <quic_ziqichen@...cinc.com>
>> ---
>> drivers/ufs/core/ufshcd.c | 15 ++++++++++++---
>> include/trace/events/ufs.h | 48 ++++++++++++++++++++++++++++++++++++++++++++++
>> 2 files changed, 60 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
>> index 3b3cf78..0037f4f 100644
>> --- a/drivers/ufs/core/ufshcd.c
>> +++ b/drivers/ufs/core/ufshcd.c
>> @@ -426,6 +426,8 @@ static void ufshcd_add_command_trace(struct ufs_hba *hba, unsigned int tag,
>> struct ufshcd_lrb *lrbp = &hba->lrb[tag];
>> struct scsi_cmnd *cmd = lrbp->cmd;
>> struct request *rq = scsi_cmd_to_rq(cmd);
>> + struct ufs_hw_queue *hwq;
>> +
>> int transfer_len = -1;
>>
>> if (!cmd)
>> @@ -456,9 +458,16 @@ static void ufshcd_add_command_trace(struct ufs_hba *hba, unsigned int tag,
>> }
>>
>> intr = ufshcd_readl(hba, REG_INTERRUPT_STATUS);
>> - doorbell = ufshcd_readl(hba, REG_UTP_TRANSFER_REQ_DOOR_BELL);
>> - trace_ufshcd_command(dev_name(hba->dev), str_t, tag,
>> - doorbell, transfer_len, intr, lba, opcode, group_id);
>> +
>> + if (is_mcq_enabled(hba)) {
>> + hwq = ufshcd_mcq_req_to_hwq(hba, rq);
>> + trace_ufshcd_command_mcq(dev_name(hba->dev), str_t, tag,
> It's better to move the processing of the "dev_name(hba->dev) into the
> trace event code. Maybe even just pass in hba.
Hi Steven,
it is a good idea, but if so do .we need to make same change to all
other UFS trace events to stay the same style.
that would be a big change.
>
>> + hwq, transfer_len, intr, lba, opcode, group_id);
>> + } else {
>> + doorbell = ufshcd_readl(hba, REG_UTP_TRANSFER_REQ_DOOR_BELL);
>> + trace_ufshcd_command(dev_name(hba->dev), str_t, tag,
> Here too.
>
>> + doorbell, transfer_len, intr, lba, opcode, group_id);
>> + }
>> }
>>
>> static void ufshcd_print_clk_freqs(struct ufs_hba *hba)
>> diff --git a/include/trace/events/ufs.h b/include/trace/events/ufs.h
>> index 599739e..a406404e 100644
>> --- a/include/trace/events/ufs.h
>> +++ b/include/trace/events/ufs.h
>> @@ -10,6 +10,7 @@
>> #define _TRACE_UFS_H
>>
>> #include <linux/tracepoint.h>
>> +#include <ufs/ufshcd.h>
>>
>> #define str_opcode(opcode) \
>> __print_symbolic(opcode, \
>> @@ -307,6 +308,53 @@ TRACE_EVENT(ufshcd_command,
>> )
>> );
>>
>> +TRACE_EVENT(ufshcd_command_mcq,
>> + TP_PROTO(const char *dev_name, enum ufs_trace_str_t str_t,
> So you would pass in the type of hba here.
>
>> + unsigned int tag, struct ufs_hw_queue *hwq, int transfer_len,
>> + u32 intr, u64 lba, u8 opcode, u8 group_id),
>> +
>> + TP_ARGS(dev_name, str_t, tag, hwq, transfer_len, intr, lba, opcode, group_id),
>> +
>> + TP_STRUCT__entry(
>> + __string(dev_name, dev_name)
> __string(dname, dev_name(hba->dev))
>
>> + __field(enum ufs_trace_str_t, str_t)
>> + __field(unsigned int, tag)
>> + __field(u32, hwq_id)
>> + __field(int, transfer_len)
>> + __field(u32, intr)
>> + __field(u64, lba)
>> + __field(u8, opcode)
>> + __field(u8, group_id)
>> + __field(u32, sq_tail)
>> + __field(u32, cq_head)
>> + __field(u32, cq_tail)
> I bet the above has a lot of holes in it. The above is not packed, so
> make sure you count the size of each field and try to keep them aligned.
you are right, I will update in next version.
Thanks，
Ziqi
>
>
>
>> + ),
>> +
>> + TP_fast_assign(
>> + __assign_str(dev_name, dev_name);
> __assign_str(dname, dev_name(hba->dev));
>
>> + __entry->str_t = str_t;
>> + __entry->tag = tag;
>> + __entry->hwq_id = hwq->id;
>> + __entry->transfer_len = transfer_len;
>> + __entry->intr = intr;
>> + __entry->lba = lba;
>> + __entry->opcode = opcode;
>> + __entry->group_id = group_id;
>> + __entry->sq_tail = hwq->sq_tail_slot;
>> + __entry->cq_head = hwq->cq_head_slot;
>> + __entry->cq_tail = hwq->cq_tail_slot;
>> + ),
>> +
>> + TP_printk(
>> + "%s: %s: tag: %u, hwq_id: %d, size: %d, IS: %u, LBA: %llu, opcode: 0x%x (%s), group_id: 0x%x, sq_tail_slot: %d, cq_head_slot: %d, cq_tail_slot: %d",
>> + show_ufs_cmd_trace_str(__entry->str_t), __get_str(dev_name),
> __get_str(dname),
>
> -- Steve
>
>> + __entry->tag, __entry->hwq_id, __entry->transfer_len,
>> + __entry->intr, __entry->lba, (u32)__entry->opcode,
>> + str_opcode(__entry->opcode), (u32)__entry->group_id,
>> + __entry->sq_tail, __entry->cq_head, __entry->cq_tail
>> + )
>> +);
>> +
>> TRACE_EVENT(ufshcd_uic_command,
>> TP_PROTO(const char *dev_name, enum ufs_trace_str_t str_t, u32 cmd,
>> u32 arg1, u32 arg2, u32 arg3),
Powered by blists - more mailing lists