Re: [PATCH] coresight: etm4x: lazily allocate memory for save_state

On 22/07/2019 21:32, Mathieu Poirier wrote:

Hi Andrew,

Sorry for the late reply - you patch got lost under the pile.

On Fri, 12 Jul 2019 at 09:01, Andrew Murray andrew.murray@arm.com wrote:

...

I had intended to lazily allocate memory for the save_state structure when it is first used. Following is a patch that I will squash into "[PATCH v3 5/6] coresight: etm4x: save/restore state across CPU low power states" on my next respin. I thought I'd share it here to get some feedback along with the rest of v3.

Thanks,

Andrew Murray

---

drivers/hwtracing/coresight/coresight-etm4x.c | 14 +++++++++++--- drivers/hwtracing/coresight/coresight-etm4x.h | 2 +- 2 files changed, 12 insertions(+), 4 deletions(-)

diff --git a/drivers/hwtracing/coresight/coresight-etm4x.c b/drivers/hwtracing/coresight/coresight-etm4x.c index b0bd8158bf13..cd02372194bc 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.c +++ b/drivers/hwtracing/coresight/coresight-etm4x.c @@ -1112,6 +1112,13 @@ static int etm4_cpu_save(struct etmv4_drvdata *drvdata) struct etmv4_save_state *state; struct device *etm_dev = &drvdata->csdev->dev;

•   if (!drvdata->save_state) {
  

•           drvdata->save_state = devm_kmalloc(etm_dev,
  

•                           sizeof(struct etmv4_save_state), GFP_KERNEL);
  

GFP_KERNEL may sleep and will not work in the context where etm4_cpu_save() is called.

Thats right and it is not worth making this GFP_ATOMIC either. We could simply decide this at probe time or when the save_restore is modified dynamically via callbacks.

Suzuki

Thanks, Mathieu

...

•           if (!drvdata->save_state)
  

•                   return -ENOMEM;
  

•   }
  

•    /*
      * As recommended by 3.4.1 ("The procedure when powering down the PE")
      * of ARM IHI 0064D
  

@@ -1134,7 +1141,7 @@ static int etm4_cpu_save(struct etmv4_drvdata *drvdata) goto out; }

•   state = &drvdata->save_state;
  

•   state = drvdata->save_state;
  
     state->trcprgctlr = readl(drvdata->base + TRCPRGCTLR);
     state->trcprocselr = readl(drvdata->base + TRCPROCSELR);
  

@@ -1234,9 +1241,10 @@ static int etm4_cpu_save(struct etmv4_drvdata *drvdata) static void etm4_cpu_restore(struct etmv4_drvdata *drvdata) { int i;

•   struct etmv4_save_state *state;
  

•   struct etmv4_save_state *state = drvdata->save_state;
  
  

•   state = &drvdata->save_state;
  

•   if (WARN_ON_ONCE(!state))
  

•           return;
  
     CS_UNLOCK(drvdata->base);
  
  

diff --git a/drivers/hwtracing/coresight/coresight-etm4x.h b/drivers/hwtracing/coresight/coresight-etm4x.h index c31634c64f87..a70cafbbb8cf 100644 --- a/drivers/hwtracing/coresight/coresight-etm4x.h +++ b/drivers/hwtracing/coresight/coresight-etm4x.h @@ -441,7 +441,7 @@ struct etmv4_drvdata { bool atbtrig; bool lpoverride; struct etmv4_config config;

•   struct etmv4_save_state         save_state;
  

•   struct etmv4_save_state         *save_state;
     bool                            state_needs_restore;
  

  };

-- 2.21.0