DPDK patches and discussions
 help / color / mirror / Atom feed
From: Andrew Boyer <aboyer@pensando.io>
To: Ferruh Yigit <ferruh.yigit@intel.com>
Cc: dev <dev@dpdk.org>, Alfredo Cardigliano <cardigliano@ntop.org>,
	Neel Patel <neel@pensando.io>
Subject: Re: [dpdk-dev] [PATCH 07/13] net/ionic: inline queue flush function
Date: Wed, 27 Jan 2021 12:43:29 -0500	[thread overview]
Message-ID: <A02135BD-1F93-4065-A466-DE54E71F23F1@pensando.io> (raw)
In-Reply-To: <1652797a-8caf-2e7c-ec25-00d4c132d603@intel.com>



> On Jan 27, 2021, at 12:36 PM, Ferruh Yigit <ferruh.yigit@intel.com> wrote:
> 
> On 1/18/2021 8:35 PM, Andrew Boyer wrote:
>> This is hot-path function.
>> Signed-off-by: Andrew Boyer <aboyer@pensando.io>
>> Signed-off-by: Neel Patel <neel@pensando.io>
>> ---
>>  drivers/net/ionic/ionic_dev.c   | 6 ------
>>  drivers/net/ionic/ionic_dev.h   | 9 ++++++++-
>>  drivers/net/ionic/ionic_osdep.h | 1 -
>>  3 files changed, 8 insertions(+), 8 deletions(-)
>> diff --git a/drivers/net/ionic/ionic_dev.c b/drivers/net/ionic/ionic_dev.c
>> index eef015686f..fcb3df482a 100644
>> --- a/drivers/net/ionic/ionic_dev.c
>> +++ b/drivers/net/ionic/ionic_dev.c
>> @@ -459,12 +459,6 @@ ionic_q_sg_map(struct ionic_queue *q, void *base, rte_iova_t base_pa)
>>  	q->sg_base_pa = base_pa;
>>  }
>>  -void
>> -ionic_q_flush(struct ionic_queue *q)
>> -{
>> -	writeq(IONIC_DBELL_QID(q->hw_index) | q->head_idx, q->db);
>> -}
>> -
>>  void
>>  ionic_q_post(struct ionic_queue *q, bool ring_doorbell, desc_cb cb,
>>  	     void *cb_arg)
>> diff --git a/drivers/net/ionic/ionic_dev.h b/drivers/net/ionic/ionic_dev.h
>> index ea89218662..55a56434fd 100644
>> --- a/drivers/net/ionic/ionic_dev.h
>> +++ b/drivers/net/ionic/ionic_dev.h
>> @@ -254,7 +254,6 @@ int ionic_q_init(struct ionic_lif *lif, struct ionic_dev *idev,
>>  	size_t desc_size, size_t sg_desc_size);
>>  void ionic_q_map(struct ionic_queue *q, void *base, rte_iova_t base_pa);
>>  void ionic_q_sg_map(struct ionic_queue *q, void *base, rte_iova_t base_pa);
>> -void ionic_q_flush(struct ionic_queue *q);
>>  void ionic_q_post(struct ionic_queue *q, bool ring_doorbell, desc_cb cb,
>>  	void *cb_arg);
>>  uint32_t ionic_q_space_avail(struct ionic_queue *q);
>> @@ -262,6 +261,14 @@ bool ionic_q_has_space(struct ionic_queue *q, uint32_t want);
>>  void ionic_q_service(struct ionic_queue *q, uint32_t cq_desc_index,
>>  	uint32_t stop_index, void *service_cb_arg);
>>  +static inline void
>> +ionic_q_flush(struct ionic_queue *q)
>> +{
>> +	uint64_t val = IONIC_DBELL_QID(q->hw_index) | q->head_idx;
>> +
>> +	rte_write64(val, q->db);
>> +}
>> +
> 
> Please be aware of '__rte_always_inline' attribute if you want to force it, since 'inline' keyword is only a suggestion to the compilers.
> 

Is it though? We are using __rte_always_inline in the hot-path .c file, but I would have said that in a header file the compiler wouldn’t have a choice. Happy to be corrected.

Thanks,
Andrew

>>  int ionic_adminq_post(struct ionic_lif *lif, struct ionic_admin_ctx *ctx);
>>    #endif /* _IONIC_DEV_H_ */
>> diff --git a/drivers/net/ionic/ionic_osdep.h b/drivers/net/ionic/ionic_osdep.h
>> index 157b0ca516..d46687b32f 100644
>> --- a/drivers/net/ionic/ionic_osdep.h
>> +++ b/drivers/net/ionic/ionic_osdep.h
>> @@ -45,6 +45,5 @@ typedef uint64_t __le64;
>>  #define ioread32(reg)		rte_read32(reg)
>>  #define iowrite8(value, reg)	rte_write8(value, reg)
>>  #define iowrite32(value, reg)	rte_write32(value, reg)
>> -#define writeq(value, reg)	rte_write64(value, reg)
>>    #endif


  reply	other threads:[~2021-01-27 17:43 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-01-18 20:34 [dpdk-dev] [PATCH 00/13] net/ionic: fixes and optimizations Andrew Boyer
2021-01-18 20:34 ` [dpdk-dev] [PATCH 01/13] net/ionic: strip out unneeded interrupt code Andrew Boyer
2021-01-18 20:34 ` [dpdk-dev] [PATCH 02/13] net/ionic: observe endianness in firmware commands Andrew Boyer
2021-01-18 20:34 ` [dpdk-dev] [PATCH 03/13] net/ionic: observe endianness in Rx filter code Andrew Boyer
2021-01-18 20:34 ` [dpdk-dev] [PATCH 04/13] net/ionic: add an array-size macro Andrew Boyer
2021-01-27 17:22   ` Ferruh Yigit
2021-01-27 17:40     ` Andrew Boyer
2021-01-29 22:44   ` [dpdk-dev] [PATCH v2 4/13] net/ionic: use the existing " Andrew Boyer
2021-02-02 12:45     ` Ferruh Yigit
2021-01-18 20:35 ` [dpdk-dev] [PATCH 05/13] net/ionic: query firmware for supported queue versions Andrew Boyer
2021-01-18 20:35 ` [dpdk-dev] [PATCH 06/13] net/ionic: clean up Tx queue version support Andrew Boyer
2021-01-27 17:30   ` Ferruh Yigit
2021-01-27 17:46     ` Andrew Boyer
2021-01-29 22:44   ` [dpdk-dev] [PATCH v2 6/13] " Andrew Boyer
2021-02-02 12:46     ` Ferruh Yigit
2021-02-05 20:20     ` Thomas Monjalon
2021-02-05 20:26       ` Andrew Boyer
2021-02-05 20:34         ` Thomas Monjalon
2021-01-18 20:35 ` [dpdk-dev] [PATCH 07/13] net/ionic: inline queue flush function Andrew Boyer
2021-01-27 17:36   ` Ferruh Yigit
2021-01-27 17:43     ` Andrew Boyer [this message]
2021-01-27 17:50       ` Ferruh Yigit
2021-01-18 20:35 ` [dpdk-dev] [PATCH 08/13] net/ionic: inline queue space function Andrew Boyer
2021-01-18 20:35 ` [dpdk-dev] [PATCH 09/13] net/ionic: observe endiannness in ioread/iowrite Andrew Boyer
2021-01-18 20:35 ` [dpdk-dev] [PATCH 10/13] net/ionic: fix to allow separate L3 and L4 csum offload Andrew Boyer
2021-01-18 20:35 ` [dpdk-dev] [PATCH 11/13] net/ionic: convert per-queue offloads into queue flags Andrew Boyer
2021-01-18 20:35 ` [dpdk-dev] [PATCH 12/13] net/ionic: fix up function attribute tags Andrew Boyer
2021-01-18 20:35 ` [dpdk-dev] [PATCH 13/13] net/ionic: fix address handling in transmit code Andrew Boyer
2021-01-27 18:02 ` [dpdk-dev] [PATCH 00/13] net/ionic: fixes and optimizations Ferruh Yigit
2021-01-27 18:10   ` Andrew Boyer
2021-01-27 22:23     ` Ferruh Yigit
2021-01-27 22:25       ` Ferruh Yigit
2021-01-29 22:44 [dpdk-dev] [PATCH] net: redefine array size macros Andrew Boyer
2021-02-01 22:28 ` Thomas Monjalon
2021-02-01 22:32   ` Andrew Boyer
2021-02-02 12:30     ` Ferruh Yigit
2021-02-22 17:09       ` Ferruh Yigit

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=A02135BD-1F93-4065-A466-DE54E71F23F1@pensando.io \
    --to=aboyer@pensando.io \
    --cc=cardigliano@ntop.org \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    --cc=neel@pensando.io \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).