| 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
| ||
|
Message-ID: <1558083934.29098.6.camel@mtksdaap41>
Date: Fri, 17 May 2019 17:05:34 +0800
From: CK Hu <ck.hu@...iatek.com>
To: Stu Hsieh <stu.hsieh@...iatek.com>
CC: Mauro Carvalho Chehab <mchehab@...nel.org>,
Rob Herring <robh+dt@...nel.org>,
Mark Rutland <mark.rutland@....com>,
"Matthias Brugger" <matthias.bgg@...il.com>,
<linux-media@...r.kernel.org>, <devicetree@...r.kernel.org>,
<linux-kernel@...r.kernel.org>,
<linux-arm-kernel@...ts.infradead.org>,
<linux-mediatek@...ts.infradead.org>, <srv_heupstream@...iatek.com>
Subject: Re: [PATCH v3 09/13] [media] mtk-mipicsi: add ISR for writing the
data to buffer
Hi, Stu:
On Tue, 2019-05-14 at 14:13 +0800, Stu Hsieh wrote:
> This patch add ISR for writing the data to buffer
>
> When mipicsi HW complete to write the data in buffer,
> the interrupt woulb be trigger.
> So, the ISR need to clear interrupt status for next interrupt.
>
> Signed-off-by: Stu Hsieh <stu.hsieh@...iatek.com>
> ---
> .../media/platform/mtk-mipicsi/mtk_mipicsi.c | 110 ++++++++++++++++++
> 1 file changed, 110 insertions(+)
>
> diff --git a/drivers/media/platform/mtk-mipicsi/mtk_mipicsi.c b/drivers/media/platform/mtk-mipicsi/mtk_mipicsi.c
> index 44c01c8d566b..af5655345754 100644
> --- a/drivers/media/platform/mtk-mipicsi/mtk_mipicsi.c
> +++ b/drivers/media/platform/mtk-mipicsi/mtk_mipicsi.c
> @@ -36,6 +36,7 @@
> #include <linux/pm_runtime.h>
> #include <linux/iommu.h>
> #include <linux/of.h>
> +#include <linux/of_irq.h>
> #include <linux/of_platform.h>
> #include <media/v4l2-common.h>
> #include <media/v4l2-dev.h>
> @@ -93,6 +94,8 @@
> #define CAMSV_MODULE_EN 0x10
> #define CAMSV_FMT_SEL 0x14
> #define CAMSV_INT_EN 0x18
> +#define CAMSV_INT_STATUS 0x1C
> +#define PASS1_DONE_STATUS 10
> #define CAMSV_SW_CTL 0x20
> #define CAMSV_CLK_EN 0x30
>
> @@ -122,6 +125,8 @@ struct mtk_mipicsi_dev {
> struct platform_device *pdev;
> unsigned int camsv_num;
> struct device *larb_pdev;
> + unsigned int irq[MTK_CAMDMA_MAX_NUM];
> + bool irq_status[MTK_CAMDMA_MAX_NUM];
> void __iomem *ana;
> void __iomem *seninf_ctrl;
> void __iomem *seninf;
> @@ -132,6 +137,7 @@ struct mtk_mipicsi_dev {
> spinlock_t lock;
> spinlock_t queue_lock;
> struct mtk_mipicsi_buf cam_buf[MAX_BUFFER_NUM];
> + bool is_enable_irq[MTK_CAMDMA_MAX_NUM];
Useless, so remove it.
> bool streamon;
> unsigned int link;
> u8 link_reg_val;
> @@ -911,9 +917,96 @@ static const struct dev_pm_ops mtk_mipicsi_pm = {
> mtk_mipicsi_pm_resume, NULL)
> };
>
> +static int get_irq_channel(struct mtk_mipicsi_dev *mipicsi)
> +{
> + int ch;
> + u32 int_reg_val;
> +
> + for (ch = 0; ch < mipicsi->camsv_num; ++ch) {
> + int_reg_val = readl(mipicsi->camsv[ch] + CAMSV_INT_STATUS);
> + if ((int_reg_val & (1UL << PASS1_DONE_STATUS)) != 0UL)
> + return ch;
> + }
> +
> + return -1;
> +}
> +
> +static void mtk_mipicsi_irq_buf_process(struct mtk_mipicsi_dev *mipicsi)
> +{
> + unsigned int i = 0U;
> + struct mtk_mipicsi_buf *new_cam_buf = NULL;
> + struct mtk_mipicsi_buf *tmp = NULL;
> + unsigned int index = 0U;
> + unsigned int next = 0U;
> +
> + for (i = 0U; i < MTK_CAMDMA_MAX_NUM; ++i)
> + mipicsi->irq_status[i] = false;
> +
> + i = 0;
> +
> + /* only one buffer left */
> + if ((&(mipicsi->fb_list))->next->next == &(mipicsi->fb_list))
> + return;
> +
> + /*for each fb_lst 2 times to get the top 2 buffer.*/
> + list_for_each_entry_safe(new_cam_buf, tmp,
> + &(mipicsi->fb_list), queue) {
> + if (i == 0U) {
> + index = new_cam_buf->vb->index;
> + } else {
> + next = new_cam_buf->vb->index;
> + break;
> + }
> + ++i;
> + }
> +
> + /*
> + * fb_list has one more buffer. Free the first buffer to user
> + * and fill the second buffer to HW.
> + */
> + vb2_buffer_done(mipicsi->cam_buf[index].vb,
> + VB2_BUF_STATE_DONE);
> +
> + list_del_init(&(mipicsi->cam_buf[index].queue));
> +}
> +
> +static irqreturn_t mtk_mipicsi_isr(int irq, void *data)
> +{
> +
> + struct mtk_mipicsi_dev *mipicsi = data;
> + unsigned long flags = 0;
> + int isr_ch;
> + u8 irq_cnt = 0, i = 0;
> +
> + spin_lock_irqsave(&mipicsi->lock, flags);
> +
> + isr_ch = get_irq_channel(mipicsi);
I think you should pass data as camsv instance, so you need not to
search the camsv instance, and each camsv instance could pointer to the
same misicsi instance.
> + if (isr_ch < 0) {
> + spin_unlock_irqrestore(&mipicsi->lock, flags);
> + return IRQ_HANDLED;
> + }
> +
> + /* clear interrupt */
> + writel(1UL << PASS1_DONE_STATUS,
> + mipicsi->camsv[isr_ch] + CAMSV_INT_STATUS);
> + mipicsi->irq_status[isr_ch] = true;
> + for (i = 0U; i < MTK_CAMDMA_MAX_NUM; ++i) {
> + if (mipicsi->irq_status[i])
> + ++irq_cnt;
> + }
> +
> + if (irq_cnt == mipicsi->link)
> + mtk_mipicsi_irq_buf_process(mipicsi);
I think mtk_mipicsi_irq_buf_process() should not be processed in irq
handler. In irq handler, do as few things as possible.
Regards,
CK
> + spin_unlock_irqrestore(&mipicsi->lock, flags);
> +
> + return IRQ_HANDLED;
> +}
> +
> static int seninf_mux_camsv_node_parse(struct mtk_mipicsi_dev *mipicsi,
> int index)
> {
> + int ret;
> + int irq;
> struct clk *clk = NULL;
> struct device *dev = NULL;
> struct resource *res = NULL;
> @@ -951,6 +1044,23 @@ static int seninf_mux_camsv_node_parse(struct mtk_mipicsi_dev *mipicsi,
> }
> mipicsi->clk[index] = clk;
>
> + irq = of_irq_get(np, 0);
> + if (irq <= 0) {
> + dev_err(dev, "get irq fail in %s node\n", np->full_name);
> + return -ENODEV;
> + }
> + mipicsi->irq[index] = irq;
> +
> + ret = devm_request_irq(dev, irq,
> + mtk_mipicsi_isr, 0,
> + mipicsi->drv_name, mipicsi);
> + if (ret != 0) {
> + dev_err(dev, "%s irq register failed\n", np->full_name);
> + return -ENODEV;
> + }
> + disable_irq(mipicsi->irq[index]);
> + mipicsi->irq_status[index] = false;
> +
> res = platform_get_resource(camdma_pdev, IORESOURCE_MEM, 0);
> if (res == NULL) {
> dev_err(dev, "get seninf_mux memory failed in %s node\n",
Powered by blists - more mailing lists