From patchwork Fri Jan 14 15:03:52 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Sean Young X-Patchwork-Id: 80007 X-Patchwork-Delegate: sean@mess.org Received: from vger.kernel.org ([23.128.96.18]) by www.linuxtv.org with esmtp (Exim 4.92) (envelope-from ) id 1n8O7U-009G1Q-QB; Fri, 14 Jan 2022 15:04:03 +0000 Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S235794AbiANPD5 (ORCPT + 1 other); Fri, 14 Jan 2022 10:03:57 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:58646 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S235747AbiANPD4 (ORCPT ); Fri, 14 Jan 2022 10:03:56 -0500 Received: from gofer.mess.org (gofer.mess.org [IPv6:2a02:8011:d000:212::1]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 55594C06161C for ; Fri, 14 Jan 2022 07:03:56 -0800 (PST) Received: by gofer.mess.org (Postfix, from userid 1000) id 7BC37101C2F; Fri, 14 Jan 2022 15:03:53 +0000 (UTC) From: Sean Young To: linux-media@vger.kernel.org, linux-mediatek@lists.infradead.org Cc: Sean Wang , Matthias Brugger Subject: [PATCH 1/2] media: mtk-cir: reduce message end to fix nec repeats Date: Fri, 14 Jan 2022 15:03:52 +0000 Message-Id: <20220114150353.195192-1-sean@mess.org> X-Mailer: git-send-email 2.30.2 MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: linux-media@vger.kernel.org X-LSpam-Score: -7.4 (-------) X-LSpam-Report: No, score=-7.4 required=5.0 tests=BAYES_00=-1.9,HEADER_FROM_DIFFERENT_DOMAINS=0.5,MAILING_LIST_MULTI=-1,RCVD_IN_DNSWL_HI=-5 autolearn=ham autolearn_force=no The ir receiver generates an interrupt with the IR data, once a space of at least ok_count is has been seen. Currently this is about 110ms; when holding down a button on a nec remote, no such space is generated until a button is released. This means nothing happens until you release the button. The sample rate is fixed at 46us, so the maximum space that can be encoded is about 12ms. So, the set ok_count above that at 23ms. Cc: Sean Wang Cc: Matthias Brugger Signed-off-by: Sean Young --- drivers/media/rc/mtk-cir.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/drivers/media/rc/mtk-cir.c b/drivers/media/rc/mtk-cir.c index 840e7aec5c21..746d43fdc17a 100644 --- a/drivers/media/rc/mtk-cir.c +++ b/drivers/media/rc/mtk-cir.c @@ -24,7 +24,8 @@ * Register to setting ok count whose unit based on hardware sampling period * indicating IR receiving completion and then making IRQ fires */ -#define MTK_OK_COUNT(x) (((x) & GENMASK(23, 16)) << 16) +#define MTK_OK_COUNT_MASK (GENMASK(22, 16)) +#define MTK_OK_COUNT(x) ((x) << 16) /* Bit to enable IR hardware function */ #define MTK_IR_EN BIT(0) @@ -268,7 +269,7 @@ static irqreturn_t mtk_ir_irq(int irqno, void *dev_id) static const struct mtk_ir_data mt7623_data = { .regs = mt7623_regs, .fields = mt7623_fields, - .ok_count = 0xf, + .ok_count = 3, .hw_period = 0xff, .div = 4, }; @@ -276,7 +277,7 @@ static const struct mtk_ir_data mt7623_data = { static const struct mtk_ir_data mt7622_data = { .regs = mt7622_regs, .fields = mt7622_fields, - .ok_count = 0xf, + .ok_count = 3, .hw_period = 0xffff, .div = 32, }; @@ -400,7 +401,7 @@ static int mtk_ir_probe(struct platform_device *pdev) mtk_w32_mask(ir, MTK_DG_CNT(1), MTK_DG_CNT_MASK, MTK_IRTHD); /* Enable IR and PWM */ - val = mtk_r32(ir, MTK_CONFIG_HIGH_REG); + val = mtk_r32(ir, MTK_CONFIG_HIGH_REG) & ~MTK_OK_COUNT_MASK; val |= MTK_OK_COUNT(ir->data->ok_count) | MTK_PWM_EN | MTK_IR_EN; mtk_w32(ir, val, MTK_CONFIG_HIGH_REG); From patchwork Fri Jan 14 15:03:53 2022 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Sean Young X-Patchwork-Id: 80008 X-Patchwork-Delegate: sean@mess.org Received: from vger.kernel.org ([23.128.96.18]) by www.linuxtv.org with esmtp (Exim 4.92) (envelope-from ) id 1n8O7Y-009G1Q-2H; Fri, 14 Jan 2022 15:04:04 +0000 Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S238287AbiANPD5 (ORCPT + 1 other); Fri, 14 Jan 2022 10:03:57 -0500 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:58644 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S235774AbiANPD4 (ORCPT ); Fri, 14 Jan 2022 10:03:56 -0500 Received: from gofer.mess.org (gofer.mess.org [IPv6:2a02:8011:d000:212::1]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 55128C061574 for ; Fri, 14 Jan 2022 07:03:56 -0800 (PST) Received: by gofer.mess.org (Postfix, from userid 1000) id 8290C101C30; Fri, 14 Jan 2022 15:03:53 +0000 (UTC) From: Sean Young To: linux-media@vger.kernel.org, linux-mediatek@lists.infradead.org Cc: Sean Wang , Matthias Brugger Subject: [PATCH 2/2] media: mtk-cir: remove superfluous ir_raw_event_reset() Date: Fri, 14 Jan 2022 15:03:53 +0000 Message-Id: <20220114150353.195192-2-sean@mess.org> X-Mailer: git-send-email 2.30.2 In-Reply-To: <20220114150353.195192-1-sean@mess.org> References: <20220114150353.195192-1-sean@mess.org> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: linux-media@vger.kernel.org X-LSpam-Score: -7.4 (-------) X-LSpam-Report: No, score=-7.4 required=5.0 tests=BAYES_00=-1.9,HEADER_FROM_DIFFERENT_DOMAINS=0.5,MAILING_LIST_MULTI=-1,RCVD_IN_DNSWL_HI=-5 autolearn=ham autolearn_force=no This IR receiver has two limitations: 1) Any IR pulse or space longer than 12ms will be truncated 12ms 2) Any pulses/spaces after the first 68 are lost ir_raw_event_reset() won't help here. If the IR cannot be decoded, any decoder should reset itself, and if it does not, this is a bug in the decoder. Cc: Sean Wang Cc: Matthias Brugger Signed-off-by: Sean Young --- drivers/media/rc/mtk-cir.c | 23 +++++++++++------------ 1 file changed, 11 insertions(+), 12 deletions(-) diff --git a/drivers/media/rc/mtk-cir.c b/drivers/media/rc/mtk-cir.c index 746d43fdc17a..255f8a116b59 100644 --- a/drivers/media/rc/mtk-cir.c +++ b/drivers/media/rc/mtk-cir.c @@ -209,19 +209,18 @@ static irqreturn_t mtk_ir_irq(int irqno, void *dev_id) struct ir_raw_event rawir = {}; /* - * Reset decoder state machine explicitly is required - * because 1) the longest duration for space MTK IR hardware - * could record is not safely long. e.g 12ms if rx resolution - * is 46us by default. There is still the risk to satisfying - * every decoder to reset themselves through long enough - * trailing spaces and 2) the IRQ handler guarantees that - * start of IR message is always contained in and starting - * from register mtk_chkdata_reg(ir, i). + * Each pulse and space is encoded as a single byte, each byte + * alternating between pulse and space. If a pulse or space is longer + * than can be encoded in a single byte, it is encoded as the maximum + * value 0xff. + * + * If a space is longer ok_count (about 23ms), the value is encoded + * as zero, and all following values are zero. Any IR that follows + * will be presented in the next interrupt. + * + * If there are more than 68 (=MTK_CHKDATA_SZ * 4) pulses and spaces, + * then the only the first 68 will be presented; the rest is lost. */ - ir_raw_event_reset(ir->rc); - - /* First message must be pulse */ - rawir.pulse = false; /* Handle all pulse and space IR controller captures */ for (i = 0 ; i < MTK_CHKDATA_SZ ; i++) {