From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from mail2.candelatech.com ([208.74.158.173]) by bombadil.infradead.org with esmtp (Exim 4.80.1 #2 (Red Hat Linux)) id 1YOB9m-00042i-PW for ath10k@lists.infradead.org; Wed, 18 Feb 2015 20:23:39 +0000 Message-ID: <54E4F4B4.4080904@candelatech.com> Date: Wed, 18 Feb 2015 12:23:16 -0800 From: Ben Greear MIME-Version: 1.0 Subject: Re: [PATCH v2] ath10k: Increase AST table SKID length limit References: <1423841776-16288-1-git-send-email-sjegadee@qti.qualcomm.com> In-Reply-To: <1423841776-16288-1-git-send-email-sjegadee@qti.qualcomm.com> List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: "ath10k" Errors-To: ath10k-bounces+kvalo=adurom.com@lists.infradead.org To: SenthilKumar Jegadeesan Cc: linux-wireless@vger.kernel.org, ath10k@lists.infradead.org Have you tried testing this with 30+ stations associated and streaming data, and with other traffic on nearby channels (say, channel 1 and 6)? We see bad performance drop-off when we set number-of-peers above 100 (with CT firmware). My kernel will set AST skid limit to number-of-peers + number-of-vdevs. Performance is best at 32 peers or less (so skid is about 40). I am wondering if the root cause is firmware/hardware walking the full skid-length for pkts from un-associated peers on same (or neighbouring) channels, or maybe unlucky hash distribution in existing peers? Thanks, Ben On 02/13/2015 07:36 AM, SenthilKumar Jegadeesan wrote: > The current SKID length configuration causes firmware > to reject peer creation for not able to allocate > AST entries for peers. This issue is observed when > least significant 3 bytes are used ramdomly to create > client MAC addresses. > > AST table SKID length configuration is increased to > maximum value to fix this issue. > > Signed-off-by: SenthilKumar Jegadeesan > --- > drivers/net/wireless/ath/ath10k/hw.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/net/wireless/ath/ath10k/hw.h b/drivers/net/wireless/ath/ath10k/hw.h > index 460771f..7f04645 100644 > --- a/drivers/net/wireless/ath/ath10k/hw.h > +++ b/drivers/net/wireless/ath/ath10k/hw.h > @@ -223,7 +223,7 @@ struct ath10k_pktlog_hdr { > #define TARGET_10X_NUM_WDS_ENTRIES 32 > #define TARGET_10X_DMA_BURST_SIZE 0 > #define TARGET_10X_MAC_AGGR_DELIM 0 > -#define TARGET_10X_AST_SKID_LIMIT 16 > +#define TARGET_10X_AST_SKID_LIMIT 128 > #define TARGET_10X_NUM_STATIONS 128 > #define TARGET_10X_NUM_PEERS ((TARGET_10X_NUM_STATIONS) + \ > (TARGET_10X_NUM_VDEVS)) > -- > 1.9.1 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-wireless" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > -- Ben Greear Candela Technologies Inc http://www.candelatech.com _______________________________________________ ath10k mailing list ath10k@lists.infradead.org http://lists.infradead.org/mailman/listinfo/ath10k From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from mail2.candelatech.com ([208.74.158.173]:56607 "EHLO mail2.candelatech.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752724AbbBRUXS (ORCPT ); Wed, 18 Feb 2015 15:23:18 -0500 Message-ID: <54E4F4B4.4080904@candelatech.com> (sfid-20150218_212336_586027_E75717E2) Date: Wed, 18 Feb 2015 12:23:16 -0800 From: Ben Greear MIME-Version: 1.0 To: SenthilKumar Jegadeesan CC: ath10k@lists.infradead.org, linux-wireless@vger.kernel.org Subject: Re: [PATCH v2] ath10k: Increase AST table SKID length limit References: <1423841776-16288-1-git-send-email-sjegadee@qti.qualcomm.com> In-Reply-To: <1423841776-16288-1-git-send-email-sjegadee@qti.qualcomm.com> Content-Type: text/plain; charset=ISO-8859-1 Sender: linux-wireless-owner@vger.kernel.org List-ID: Have you tried testing this with 30+ stations associated and streaming data, and with other traffic on nearby channels (say, channel 1 and 6)? We see bad performance drop-off when we set number-of-peers above 100 (with CT firmware). My kernel will set AST skid limit to number-of-peers + number-of-vdevs. Performance is best at 32 peers or less (so skid is about 40). I am wondering if the root cause is firmware/hardware walking the full skid-length for pkts from un-associated peers on same (or neighbouring) channels, or maybe unlucky hash distribution in existing peers? Thanks, Ben On 02/13/2015 07:36 AM, SenthilKumar Jegadeesan wrote: > The current SKID length configuration causes firmware > to reject peer creation for not able to allocate > AST entries for peers. This issue is observed when > least significant 3 bytes are used ramdomly to create > client MAC addresses. > > AST table SKID length configuration is increased to > maximum value to fix this issue. > > Signed-off-by: SenthilKumar Jegadeesan > --- > drivers/net/wireless/ath/ath10k/hw.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/net/wireless/ath/ath10k/hw.h b/drivers/net/wireless/ath/ath10k/hw.h > index 460771f..7f04645 100644 > --- a/drivers/net/wireless/ath/ath10k/hw.h > +++ b/drivers/net/wireless/ath/ath10k/hw.h > @@ -223,7 +223,7 @@ struct ath10k_pktlog_hdr { > #define TARGET_10X_NUM_WDS_ENTRIES 32 > #define TARGET_10X_DMA_BURST_SIZE 0 > #define TARGET_10X_MAC_AGGR_DELIM 0 > -#define TARGET_10X_AST_SKID_LIMIT 16 > +#define TARGET_10X_AST_SKID_LIMIT 128 > #define TARGET_10X_NUM_STATIONS 128 > #define TARGET_10X_NUM_PEERS ((TARGET_10X_NUM_STATIONS) + \ > (TARGET_10X_NUM_VDEVS)) > -- > 1.9.1 > > -- > To unsubscribe from this list: send the line "unsubscribe linux-wireless" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html > -- Ben Greear Candela Technologies Inc http://www.candelatech.com