linux-kernel - Re: [PATCH 4/6] scsi: pm8001: Use sas_task_find

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

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <1303aefc-6a06-371d-5fc0-828bbce29ad4@opensource.wdc.com>
Date:   Thu, 29 Sep 2022 11:17:41 +0900
From:   Damien Le Moal <damien.lemoal@...nsource.wdc.com>
To:     John Garry <john.garry@...wei.com>, jejb@...ux.ibm.com,
        martin.petersen@...cle.com, jinpu.wang@...ud.ionos.com,
        damien.lemoal@....com
Cc:     hare@...e.de, linux-scsi@...r.kernel.org,
        linux-kernel@...r.kernel.org, linuxarm@...wei.com,
        ipylypiv@...gle.com, changyuanl@...gle.com, hch@....de
Subject: Re: [PATCH 4/6] scsi: pm8001: Use sas_task_find_rq() for tagging

On 9/28/22 21:27, John Garry wrote:
> The request associated with a scsi command coming from the block layer
> has a unique tag, so use that when possible for getting a CCB.
> 
> Unfortunately we don't support reserved commands in the SCSI midlayer yet,
> so in the interim continue to manage those tags internally (along with
> tags for private commands).
> 
> Signed-off-by: John Garry <john.garry@...wei.com>
> ---
>  drivers/scsi/pm8001/pm8001_init.c | 10 ++++------
>  drivers/scsi/pm8001/pm8001_sas.c  |  8 ++++++++
>  drivers/scsi/pm8001/pm8001_sas.h  |  6 +++++-
>  3 files changed, 17 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/scsi/pm8001/pm8001_init.c b/drivers/scsi/pm8001/pm8001_init.c
> index 0edc9857a8bd..0868836e7391 100644
> --- a/drivers/scsi/pm8001/pm8001_init.c
> +++ b/drivers/scsi/pm8001/pm8001_init.c
> @@ -1208,17 +1208,14 @@ static int pm8001_init_ccb_tag(struct pm8001_hba_info *pm8001_ha)
>  	struct Scsi_Host *shost = pm8001_ha->shost;
>  	struct device *dev = pm8001_ha->dev;
>  	u32 max_out_io, ccb_count;
> -	u32 can_queue;
>  	int i;
>  
>  	max_out_io = pm8001_ha->main_cfg_tbl.pm80xx_tbl.max_out_io;
>  	ccb_count = min_t(int, PM8001_MAX_CCB, max_out_io);
>  
> -	/* Update to the scsi host*/
> -	can_queue = ccb_count - PM8001_RESERVE_SLOT;
> -	shost->can_queue = can_queue;
> +	shost->can_queue = ccb_count - PM8001_RESERVE_SLOT;
>  
> -	pm8001_ha->tags = bitmap_zalloc(ccb_count, GFP_KERNEL);
> +	pm8001_ha->tags = bitmap_zalloc(PM8001_RESERVE_SLOT, GFP_KERNEL);

The "tags" name for this field is really confusing as it seems to be
implying "all tags". Could we rename that to reserved_tags or similar ?

>  	if (!pm8001_ha->tags)
>  		goto err_out;
>  
> @@ -1244,9 +1241,10 @@ static int pm8001_init_ccb_tag(struct pm8001_hba_info *pm8001_ha)
>  		pm8001_ha->ccb_info[i].task = NULL;
>  		pm8001_ha->ccb_info[i].ccb_tag = PM8001_INVALID_TAG;
>  		pm8001_ha->ccb_info[i].device = NULL;
> -		++pm8001_ha->tags_num;
>  	}
>  
> +	pm8001_ha->tags_num = PM8001_RESERVE_SLOT;

Same here. reserved_tags_num ?
But given that this seems to always be equal to PM8001_RESERVE_SLOT, do
we even need this field at all ?

> +
>  	return 0;
>  
>  err_out_noccb:
> diff --git a/drivers/scsi/pm8001/pm8001_sas.c b/drivers/scsi/pm8001/pm8001_sas.c
> index 066dfa9f4683..9d25855af657 100644
> --- a/drivers/scsi/pm8001/pm8001_sas.c
> +++ b/drivers/scsi/pm8001/pm8001_sas.c
> @@ -68,6 +68,11 @@ void pm8001_tag_free(struct pm8001_hba_info *pm8001_ha, u32 tag)
>  	void *bitmap = pm8001_ha->tags;
>  	unsigned long flags;
>  
> +	if (tag < pm8001_ha->shost->can_queue)
> +		return;
> +
> +	tag -= pm8001_ha->shost->can_queue;
> +
>  	spin_lock_irqsave(&pm8001_ha->bitmap_lock, flags);
>  	__clear_bit(tag, bitmap);
>  	spin_unlock_irqrestore(&pm8001_ha->bitmap_lock, flags);
> @@ -92,6 +97,9 @@ int pm8001_tag_alloc(struct pm8001_hba_info *pm8001_ha, u32 *tag_out)
>  	}
>  	__set_bit(tag, bitmap);
>  	spin_unlock_irqrestore(&pm8001_ha->bitmap_lock, flags);
> +
> +	/* reserved tags are in the upper region of the tagset */
> +	tag += pm8001_ha->shost->can_queue;
>  	*tag_out = tag;
>  	return 0;
>  }
> diff --git a/drivers/scsi/pm8001/pm8001_sas.h b/drivers/scsi/pm8001/pm8001_sas.h
> index 9acaadf02150..9ff8d1fa84b0 100644
> --- a/drivers/scsi/pm8001/pm8001_sas.h
> +++ b/drivers/scsi/pm8001/pm8001_sas.h
> @@ -737,9 +737,13 @@ pm8001_ccb_alloc(struct pm8001_hba_info *pm8001_ha,
>  		 struct pm8001_device *dev, struct sas_task *task)
>  {
>  	struct pm8001_ccb_info *ccb;
> +	struct request *rq = NULL;

I do not think you need the NULL initialization...

>  	u32 tag;
>  
> -	if (pm8001_tag_alloc(pm8001_ha, &tag)) {
> +	rq = sas_task_find_rq(task);
> +	if (rq) {
> +		tag = rq->tag;
> +	} else if (pm8001_tag_alloc(pm8001_ha, &tag)) {
>  		pm8001_dbg(pm8001_ha, FAIL, "Failed to allocate a tag\n");
>  		return NULL;
>  	}

-- 
Damien Le Moal
Western Digital Research