* [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"
@ 2025-02-08 22:30 Caleb Connolly
2025-02-08 22:33 ` Caleb Connolly
` (4 more replies)
0 siblings, 5 replies; 9+ messages in thread
From: Caleb Connolly @ 2025-02-08 22:30 UTC (permalink / raw)
To: Uwe Kleine-König, Amit Vadhavana, Caleb Connolly, Dave Jiang,
Fenghua Yu, Kees Cook, Md Sadre Alam, Robin Murphy, Vinod Koul
Cc: David Heidelberg, Dmitry Baryshkov, dmaengine, linux-arm-msm
This commit causes a hard crash on sdm845 and likely other platforms.
Revert it until a proper fix is found.
This reverts commit 57a7138d0627309d469719f1845d2778c251f358.
Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
---
drivers/dma/qcom/bam_dma.c | 24 ++++++++----------------
1 file changed, 8 insertions(+), 16 deletions(-)
diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c
index c14557efd577..bbc3276992bb 100644
--- a/drivers/dma/qcom/bam_dma.c
+++ b/drivers/dma/qcom/bam_dma.c
@@ -58,11 +58,8 @@ struct bam_desc_hw {
#define DESC_FLAG_EOB BIT(13)
#define DESC_FLAG_NWD BIT(12)
#define DESC_FLAG_CMD BIT(11)
-#define BAM_NDP_REVISION_START 0x20
-#define BAM_NDP_REVISION_END 0x27
-
struct bam_async_desc {
struct virt_dma_desc vd;
u32 num_desc;
@@ -400,9 +397,8 @@ struct bam_device {
int irq;
/* dma start transaction tasklet */
struct tasklet_struct task;
- u32 bam_revision;
};
/**
* bam_addr - returns BAM register address
@@ -444,12 +440,10 @@ static void bam_reset(struct bam_device *bdev)
val |= BAM_EN;
writel_relaxed(val, bam_addr(bdev, 0, BAM_CTRL));
/* set descriptor threshold, start with 4 bytes */
- if (in_range(bdev->bam_revision, BAM_NDP_REVISION_START,
- BAM_NDP_REVISION_END))
- writel_relaxed(DEFAULT_CNT_THRSHLD,
- bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
+ writel_relaxed(DEFAULT_CNT_THRSHLD,
+ bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
/* Enable default set of h/w workarounds, ie all except BAM_FULL_PIPE */
writel_relaxed(BAM_CNFG_BITS_DEFAULT, bam_addr(bdev, 0, BAM_CNFG_BITS));
@@ -1005,12 +999,11 @@ static void bam_apply_new_config(struct bam_chan *bchan,
if (dir == DMA_DEV_TO_MEM)
maxburst = bchan->slave.src_maxburst;
else
maxburst = bchan->slave.dst_maxburst;
- if (in_range(bdev->bam_revision, BAM_NDP_REVISION_START,
- BAM_NDP_REVISION_END))
- writel_relaxed(maxburst,
- bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
+
+ writel_relaxed(maxburst,
+ bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
}
bchan->reconfigure = 0;
}
@@ -1198,13 +1191,12 @@ static int bam_init(struct bam_device *bdev)
{
u32 val;
/* read revision and configuration information */
- val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
- if (!bdev->num_ees)
+ if (!bdev->num_ees) {
+ val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
bdev->num_ees = (val >> NUM_EES_SHIFT) & NUM_EES_MASK;
-
- bdev->bam_revision = val & REVISION_MASK;
+ }
/* check that configured EE is within range */
if (bdev->ee >= bdev->num_ees)
return -EINVAL;
--
2.48.1
^ permalink raw reply related [flat|nested] 9+ messages in thread
* Re: [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"
2025-02-08 22:30 [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register" Caleb Connolly
@ 2025-02-08 22:33 ` Caleb Connolly
2025-02-09 0:23 ` David Heidelberg
` (3 subsequent siblings)
4 siblings, 0 replies; 9+ messages in thread
From: Caleb Connolly @ 2025-02-08 22:33 UTC (permalink / raw)
To: Uwe Kleine-König, Amit Vadhavana, Dave Jiang, Kees Cook,
Md Sadre Alam, Robin Murphy, Vinod Koul
Cc: David Heidelberg, Dmitry Baryshkov, dmaengine, linux-arm-msm
On 2/8/25 22:30, Caleb Connolly wrote:
> This commit causes a hard crash on sdm845 and likely other platforms.
> Revert it until a proper fix is found.
>
> This reverts commit 57a7138d0627309d469719f1845d2778c251f358.
>
> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
Missing fixes tag
Fixes: 57a7138d0627 (dmaengine: qcom: bam_dma: Avoid writing unavailable
register)
> ---
> drivers/dma/qcom/bam_dma.c | 24 ++++++++----------------
> 1 file changed, 8 insertions(+), 16 deletions(-)
>
> diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c
> index c14557efd577..bbc3276992bb 100644
> --- a/drivers/dma/qcom/bam_dma.c
> +++ b/drivers/dma/qcom/bam_dma.c
> @@ -58,11 +58,8 @@ struct bam_desc_hw {
> #define DESC_FLAG_EOB BIT(13)
> #define DESC_FLAG_NWD BIT(12)
> #define DESC_FLAG_CMD BIT(11)
>
> -#define BAM_NDP_REVISION_START 0x20
> -#define BAM_NDP_REVISION_END 0x27
> -
> struct bam_async_desc {
> struct virt_dma_desc vd;
>
> u32 num_desc;
> @@ -400,9 +397,8 @@ struct bam_device {
> int irq;
>
> /* dma start transaction tasklet */
> struct tasklet_struct task;
> - u32 bam_revision;
> };
>
> /**
> * bam_addr - returns BAM register address
> @@ -444,12 +440,10 @@ static void bam_reset(struct bam_device *bdev)
> val |= BAM_EN;
> writel_relaxed(val, bam_addr(bdev, 0, BAM_CTRL));
>
> /* set descriptor threshold, start with 4 bytes */
> - if (in_range(bdev->bam_revision, BAM_NDP_REVISION_START,
> - BAM_NDP_REVISION_END))
> - writel_relaxed(DEFAULT_CNT_THRSHLD,
> - bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> + writel_relaxed(DEFAULT_CNT_THRSHLD,
> + bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
>
> /* Enable default set of h/w workarounds, ie all except BAM_FULL_PIPE */
> writel_relaxed(BAM_CNFG_BITS_DEFAULT, bam_addr(bdev, 0, BAM_CNFG_BITS));
>
> @@ -1005,12 +999,11 @@ static void bam_apply_new_config(struct bam_chan *bchan,
> if (dir == DMA_DEV_TO_MEM)
> maxburst = bchan->slave.src_maxburst;
> else
> maxburst = bchan->slave.dst_maxburst;
> - if (in_range(bdev->bam_revision, BAM_NDP_REVISION_START,
> - BAM_NDP_REVISION_END))
> - writel_relaxed(maxburst,
> - bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> +
> + writel_relaxed(maxburst,
> + bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> }
>
> bchan->reconfigure = 0;
> }
> @@ -1198,13 +1191,12 @@ static int bam_init(struct bam_device *bdev)
> {
> u32 val;
>
> /* read revision and configuration information */
> - val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
> - if (!bdev->num_ees)
> + if (!bdev->num_ees) {
> + val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
> bdev->num_ees = (val >> NUM_EES_SHIFT) & NUM_EES_MASK;
> -
> - bdev->bam_revision = val & REVISION_MASK;
> + }
>
> /* check that configured EE is within range */
> if (bdev->ee >= bdev->num_ees)
> return -EINVAL;
--
Caleb (they/them)
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"
2025-02-08 22:30 [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register" Caleb Connolly
2025-02-08 22:33 ` Caleb Connolly
@ 2025-02-09 0:23 ` David Heidelberg
2025-02-09 2:03 ` Bjorn Andersson
` (2 subsequent siblings)
4 siblings, 0 replies; 9+ messages in thread
From: David Heidelberg @ 2025-02-09 0:23 UTC (permalink / raw)
To: Caleb Connolly, Uwe Kleine-König, Amit Vadhavana, Dave Jiang,
Fenghua Yu, Kees Cook, Md Sadre Alam, Robin Murphy, Vinod Koul
Cc: Dmitry Baryshkov, dmaengine, linux-arm-msm
Fixes initial boot crash into QCOM Crashdump.
Tested-by: David Heidelberg <david@ixit.cz>
On 08/02/2025 23:30, Caleb Connolly wrote:
> This commit causes a hard crash on sdm845 and likely other platforms.
> Revert it until a proper fix is found.
>
> This reverts commit 57a7138d0627309d469719f1845d2778c251f358.
>
> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
> ---
> drivers/dma/qcom/bam_dma.c | 24 ++++++++----------------
> 1 file changed, 8 insertions(+), 16 deletions(-)
>
> diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c
> index c14557efd577..bbc3276992bb 100644
> --- a/drivers/dma/qcom/bam_dma.c
> +++ b/drivers/dma/qcom/bam_dma.c
> @@ -58,11 +58,8 @@ struct bam_desc_hw {
> #define DESC_FLAG_EOB BIT(13)
> #define DESC_FLAG_NWD BIT(12)
> #define DESC_FLAG_CMD BIT(11)
>
> -#define BAM_NDP_REVISION_START 0x20
> -#define BAM_NDP_REVISION_END 0x27
> -
> struct bam_async_desc {
> struct virt_dma_desc vd;
>
> u32 num_desc;
> @@ -400,9 +397,8 @@ struct bam_device {
> int irq;
>
> /* dma start transaction tasklet */
> struct tasklet_struct task;
> - u32 bam_revision;
> };
>
> /**
> * bam_addr - returns BAM register address
> @@ -444,12 +440,10 @@ static void bam_reset(struct bam_device *bdev)
> val |= BAM_EN;
> writel_relaxed(val, bam_addr(bdev, 0, BAM_CTRL));
>
> /* set descriptor threshold, start with 4 bytes */
> - if (in_range(bdev->bam_revision, BAM_NDP_REVISION_START,
> - BAM_NDP_REVISION_END))
> - writel_relaxed(DEFAULT_CNT_THRSHLD,
> - bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> + writel_relaxed(DEFAULT_CNT_THRSHLD,
> + bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
>
> /* Enable default set of h/w workarounds, ie all except BAM_FULL_PIPE */
> writel_relaxed(BAM_CNFG_BITS_DEFAULT, bam_addr(bdev, 0, BAM_CNFG_BITS));
>
> @@ -1005,12 +999,11 @@ static void bam_apply_new_config(struct bam_chan *bchan,
> if (dir == DMA_DEV_TO_MEM)
> maxburst = bchan->slave.src_maxburst;
> else
> maxburst = bchan->slave.dst_maxburst;
> - if (in_range(bdev->bam_revision, BAM_NDP_REVISION_START,
> - BAM_NDP_REVISION_END))
> - writel_relaxed(maxburst,
> - bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> +
> + writel_relaxed(maxburst,
> + bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> }
>
> bchan->reconfigure = 0;
> }
> @@ -1198,13 +1191,12 @@ static int bam_init(struct bam_device *bdev)
> {
> u32 val;
>
> /* read revision and configuration information */
> - val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
> - if (!bdev->num_ees)
> + if (!bdev->num_ees) {
> + val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
> bdev->num_ees = (val >> NUM_EES_SHIFT) & NUM_EES_MASK;
> -
> - bdev->bam_revision = val & REVISION_MASK;
> + }
>
> /* check that configured EE is within range */
> if (bdev->ee >= bdev->num_ees)
> return -EINVAL;
--
David Heidelberg
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"
2025-02-08 22:30 [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register" Caleb Connolly
2025-02-08 22:33 ` Caleb Connolly
2025-02-09 0:23 ` David Heidelberg
@ 2025-02-09 2:03 ` Bjorn Andersson
2025-02-10 10:19 ` Vinod Koul
2025-02-10 11:39 ` neil.armstrong
2025-02-27 12:07 ` Vinod Koul
4 siblings, 1 reply; 9+ messages in thread
From: Bjorn Andersson @ 2025-02-09 2:03 UTC (permalink / raw)
To: Caleb Connolly
Cc: Uwe Kleine-König, Amit Vadhavana, Dave Jiang, Fenghua Yu,
Kees Cook, Md Sadre Alam, Robin Murphy, Vinod Koul,
David Heidelberg, Dmitry Baryshkov, dmaengine, linux-arm-msm
On Sat, Feb 08, 2025 at 10:30:54PM +0000, Caleb Connolly wrote:
> This commit causes a hard crash on sdm845 and likely other platforms.
> Revert it until a proper fix is found.
>
> This reverts commit 57a7138d0627309d469719f1845d2778c251f358.
>
I posted below patch yesterday, which reverts the change for
bdev->num_ees != 0 (i.e. SDM845), while still retaining the introduced
NDP vs Lite logic.
https://lore.kernel.org/linux-arm-msm/0892dca2-e76b-4aab-95cf-7437dabfc7a4@kernel.org/T/#t
Regards,
Bjorn
> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
> ---
> drivers/dma/qcom/bam_dma.c | 24 ++++++++----------------
> 1 file changed, 8 insertions(+), 16 deletions(-)
>
> diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c
> index c14557efd577..bbc3276992bb 100644
> --- a/drivers/dma/qcom/bam_dma.c
> +++ b/drivers/dma/qcom/bam_dma.c
> @@ -58,11 +58,8 @@ struct bam_desc_hw {
> #define DESC_FLAG_EOB BIT(13)
> #define DESC_FLAG_NWD BIT(12)
> #define DESC_FLAG_CMD BIT(11)
>
> -#define BAM_NDP_REVISION_START 0x20
> -#define BAM_NDP_REVISION_END 0x27
> -
> struct bam_async_desc {
> struct virt_dma_desc vd;
>
> u32 num_desc;
> @@ -400,9 +397,8 @@ struct bam_device {
> int irq;
>
> /* dma start transaction tasklet */
> struct tasklet_struct task;
> - u32 bam_revision;
> };
>
> /**
> * bam_addr - returns BAM register address
> @@ -444,12 +440,10 @@ static void bam_reset(struct bam_device *bdev)
> val |= BAM_EN;
> writel_relaxed(val, bam_addr(bdev, 0, BAM_CTRL));
>
> /* set descriptor threshold, start with 4 bytes */
> - if (in_range(bdev->bam_revision, BAM_NDP_REVISION_START,
> - BAM_NDP_REVISION_END))
> - writel_relaxed(DEFAULT_CNT_THRSHLD,
> - bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> + writel_relaxed(DEFAULT_CNT_THRSHLD,
> + bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
>
> /* Enable default set of h/w workarounds, ie all except BAM_FULL_PIPE */
> writel_relaxed(BAM_CNFG_BITS_DEFAULT, bam_addr(bdev, 0, BAM_CNFG_BITS));
>
> @@ -1005,12 +999,11 @@ static void bam_apply_new_config(struct bam_chan *bchan,
> if (dir == DMA_DEV_TO_MEM)
> maxburst = bchan->slave.src_maxburst;
> else
> maxburst = bchan->slave.dst_maxburst;
> - if (in_range(bdev->bam_revision, BAM_NDP_REVISION_START,
> - BAM_NDP_REVISION_END))
> - writel_relaxed(maxburst,
> - bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> +
> + writel_relaxed(maxburst,
> + bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> }
>
> bchan->reconfigure = 0;
> }
> @@ -1198,13 +1191,12 @@ static int bam_init(struct bam_device *bdev)
> {
> u32 val;
>
> /* read revision and configuration information */
> - val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
> - if (!bdev->num_ees)
> + if (!bdev->num_ees) {
> + val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
> bdev->num_ees = (val >> NUM_EES_SHIFT) & NUM_EES_MASK;
> -
> - bdev->bam_revision = val & REVISION_MASK;
> + }
>
> /* check that configured EE is within range */
> if (bdev->ee >= bdev->num_ees)
> return -EINVAL;
> --
> 2.48.1
>
>
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"
2025-02-09 2:03 ` Bjorn Andersson
@ 2025-02-10 10:19 ` Vinod Koul
2025-02-10 11:40 ` neil.armstrong
2025-02-24 13:16 ` Manivannan Sadhasivam
0 siblings, 2 replies; 9+ messages in thread
From: Vinod Koul @ 2025-02-10 10:19 UTC (permalink / raw)
To: Bjorn Andersson
Cc: Caleb Connolly, Uwe Kleine-König, Amit Vadhavana, Dave Jiang,
Fenghua Yu, Kees Cook, Md Sadre Alam, Robin Murphy,
David Heidelberg, Dmitry Baryshkov, dmaengine, linux-arm-msm
On 08-02-25, 20:03, Bjorn Andersson wrote:
> On Sat, Feb 08, 2025 at 10:30:54PM +0000, Caleb Connolly wrote:
> > This commit causes a hard crash on sdm845 and likely other platforms.
> > Revert it until a proper fix is found.
> >
> > This reverts commit 57a7138d0627309d469719f1845d2778c251f358.
> >
>
> I posted below patch yesterday, which reverts the change for
> bdev->num_ees != 0 (i.e. SDM845), while still retaining the introduced
> NDP vs Lite logic.
Bjorn,
Given the issues reported by Stephan, do you this we should revert
and then patch it up..?
--
~Vinod
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"
2025-02-08 22:30 [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register" Caleb Connolly
` (2 preceding siblings ...)
2025-02-09 2:03 ` Bjorn Andersson
@ 2025-02-10 11:39 ` neil.armstrong
2025-02-27 12:07 ` Vinod Koul
4 siblings, 0 replies; 9+ messages in thread
From: neil.armstrong @ 2025-02-10 11:39 UTC (permalink / raw)
To: Caleb Connolly, Uwe Kleine-König, Amit Vadhavana, Dave Jiang,
Fenghua Yu, Kees Cook, Md Sadre Alam, Robin Murphy, Vinod Koul
Cc: David Heidelberg, Dmitry Baryshkov, dmaengine, linux-arm-msm
On 08/02/2025 23:30, Caleb Connolly wrote:
> This commit causes a hard crash on sdm845 and likely other platforms.
> Revert it until a proper fix is found.
>
> This reverts commit 57a7138d0627309d469719f1845d2778c251f358.
>
> Signed-off-by: Caleb Connolly <caleb.connolly@linaro.org>
> ---
> drivers/dma/qcom/bam_dma.c | 24 ++++++++----------------
> 1 file changed, 8 insertions(+), 16 deletions(-)
>
> diff --git a/drivers/dma/qcom/bam_dma.c b/drivers/dma/qcom/bam_dma.c
> index c14557efd577..bbc3276992bb 100644
> --- a/drivers/dma/qcom/bam_dma.c
> +++ b/drivers/dma/qcom/bam_dma.c
> @@ -58,11 +58,8 @@ struct bam_desc_hw {
> #define DESC_FLAG_EOB BIT(13)
> #define DESC_FLAG_NWD BIT(12)
> #define DESC_FLAG_CMD BIT(11)
>
> -#define BAM_NDP_REVISION_START 0x20
> -#define BAM_NDP_REVISION_END 0x27
> -
> struct bam_async_desc {
> struct virt_dma_desc vd;
>
> u32 num_desc;
> @@ -400,9 +397,8 @@ struct bam_device {
> int irq;
>
> /* dma start transaction tasklet */
> struct tasklet_struct task;
> - u32 bam_revision;
> };
>
> /**
> * bam_addr - returns BAM register address
> @@ -444,12 +440,10 @@ static void bam_reset(struct bam_device *bdev)
> val |= BAM_EN;
> writel_relaxed(val, bam_addr(bdev, 0, BAM_CTRL));
>
> /* set descriptor threshold, start with 4 bytes */
> - if (in_range(bdev->bam_revision, BAM_NDP_REVISION_START,
> - BAM_NDP_REVISION_END))
> - writel_relaxed(DEFAULT_CNT_THRSHLD,
> - bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> + writel_relaxed(DEFAULT_CNT_THRSHLD,
> + bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
>
> /* Enable default set of h/w workarounds, ie all except BAM_FULL_PIPE */
> writel_relaxed(BAM_CNFG_BITS_DEFAULT, bam_addr(bdev, 0, BAM_CNFG_BITS));
>
> @@ -1005,12 +999,11 @@ static void bam_apply_new_config(struct bam_chan *bchan,
> if (dir == DMA_DEV_TO_MEM)
> maxburst = bchan->slave.src_maxburst;
> else
> maxburst = bchan->slave.dst_maxburst;
> - if (in_range(bdev->bam_revision, BAM_NDP_REVISION_START,
> - BAM_NDP_REVISION_END))
> - writel_relaxed(maxburst,
> - bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> +
> + writel_relaxed(maxburst,
> + bam_addr(bdev, 0, BAM_DESC_CNT_TRSHLD));
> }
>
> bchan->reconfigure = 0;
> }
> @@ -1198,13 +1191,12 @@ static int bam_init(struct bam_device *bdev)
> {
> u32 val;
>
> /* read revision and configuration information */
> - val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
> - if (!bdev->num_ees)
> + if (!bdev->num_ees) {
> + val = readl_relaxed(bam_addr(bdev, 0, BAM_REVISION));
> bdev->num_ees = (val >> NUM_EES_SHIFT) & NUM_EES_MASK;
> -
> - bdev->bam_revision = val & REVISION_MASK;
> + }
>
> /* check that configured EE is within range */
> if (bdev->ee >= bdev->num_ees)
> return -EINVAL;
This fixes the BAM DMA crash happening on DB845c & ifc6410.
Tested-by: Neil Armstrong <neil.armstrong@linaro.org> # on sdm845-DB845c
Tested-by: Neil Armstrong <neil.armstrong@linaro.org> # on apq8064-ifc6410
CI: https://git.codelinaro.org/linaro/qcomlt/ci/staging/linux/-/pipelines/131277
Thanks,
Neil
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"
2025-02-10 10:19 ` Vinod Koul
@ 2025-02-10 11:40 ` neil.armstrong
2025-02-24 13:16 ` Manivannan Sadhasivam
1 sibling, 0 replies; 9+ messages in thread
From: neil.armstrong @ 2025-02-10 11:40 UTC (permalink / raw)
To: Vinod Koul, Bjorn Andersson
Cc: Caleb Connolly, Uwe Kleine-König, Amit Vadhavana, Dave Jiang,
Fenghua Yu, Kees Cook, Md Sadre Alam, Robin Murphy,
David Heidelberg, Dmitry Baryshkov, dmaengine, linux-arm-msm
On 10/02/2025 11:19, Vinod Koul wrote:
> On 08-02-25, 20:03, Bjorn Andersson wrote:
>> On Sat, Feb 08, 2025 at 10:30:54PM +0000, Caleb Connolly wrote:
>>> This commit causes a hard crash on sdm845 and likely other platforms.
>>> Revert it until a proper fix is found.
>>>
>>> This reverts commit 57a7138d0627309d469719f1845d2778c251f358.
>>>
>>
>> I posted below patch yesterday, which reverts the change for
>> bdev->num_ees != 0 (i.e. SDM845), while still retaining the introduced
>> NDP vs Lite logic.
>
> Bjorn,
>
> Given the issues reported by Stephan, do you this we should revert
> and then patch it up..?
>
With Bjorn's change, DB845c is fixed, but ifc6410 is still broken:
https://git.codelinaro.org/linaro/qcomlt/ci/staging/cdba-tester/-/pipelines/131282
So please revert to it can be resubmitted correctly
Neil
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"
2025-02-10 10:19 ` Vinod Koul
2025-02-10 11:40 ` neil.armstrong
@ 2025-02-24 13:16 ` Manivannan Sadhasivam
1 sibling, 0 replies; 9+ messages in thread
From: Manivannan Sadhasivam @ 2025-02-24 13:16 UTC (permalink / raw)
To: Vinod Koul
Cc: Bjorn Andersson, Caleb Connolly, Uwe Kleine-König,
Amit Vadhavana, Dave Jiang, Fenghua Yu, Kees Cook, Md Sadre Alam,
Robin Murphy, David Heidelberg, Dmitry Baryshkov, dmaengine,
linux-arm-msm
On Mon, Feb 10, 2025 at 03:49:52PM +0530, Vinod Koul wrote:
> On 08-02-25, 20:03, Bjorn Andersson wrote:
> > On Sat, Feb 08, 2025 at 10:30:54PM +0000, Caleb Connolly wrote:
> > > This commit causes a hard crash on sdm845 and likely other platforms.
> > > Revert it until a proper fix is found.
> > >
> > > This reverts commit 57a7138d0627309d469719f1845d2778c251f358.
> > >
> >
> > I posted below patch yesterday, which reverts the change for
> > bdev->num_ees != 0 (i.e. SDM845), while still retaining the introduced
> > NDP vs Lite logic.
>
> Bjorn,
>
> Given the issues reported by Stephan, do you this we should revert
> and then patch it up..?
>
Vinod, please apply this revert patch as there doesn't seem to be a proper fix.
The offending commit causes hang in multiple platforms.
- Mani
--
மணிவண்ணன் சதாசிவம்
^ permalink raw reply [flat|nested] 9+ messages in thread
* Re: [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"
2025-02-08 22:30 [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register" Caleb Connolly
` (3 preceding siblings ...)
2025-02-10 11:39 ` neil.armstrong
@ 2025-02-27 12:07 ` Vinod Koul
4 siblings, 0 replies; 9+ messages in thread
From: Vinod Koul @ 2025-02-27 12:07 UTC (permalink / raw)
To: Uwe Kleine-König, Amit Vadhavana, Dave Jiang, Fenghua Yu,
Kees Cook, Md Sadre Alam, Robin Murphy, Caleb Connolly
Cc: David Heidelberg, Dmitry Baryshkov, dmaengine, linux-arm-msm
On Sat, 08 Feb 2025 22:30:54 +0000, Caleb Connolly wrote:
> This commit causes a hard crash on sdm845 and likely other platforms.
> Revert it until a proper fix is found.
>
> This reverts commit 57a7138d0627309d469719f1845d2778c251f358.
>
>
Applied, thanks!
[1/1] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"
commit: e521f516716de7895acd1b5b7fac788214a390b9
Best regards,
--
~Vinod
^ permalink raw reply [flat|nested] 9+ messages in thread
end of thread, other threads:[~2025-02-27 12:07 UTC | newest]
Thread overview: 9+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-02-08 22:30 [PATCH] Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register" Caleb Connolly
2025-02-08 22:33 ` Caleb Connolly
2025-02-09 0:23 ` David Heidelberg
2025-02-09 2:03 ` Bjorn Andersson
2025-02-10 10:19 ` Vinod Koul
2025-02-10 11:40 ` neil.armstrong
2025-02-24 13:16 ` Manivannan Sadhasivam
2025-02-10 11:39 ` neil.armstrong
2025-02-27 12:07 ` Vinod Koul
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox