[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 23 Feb 2021 18:40:26 +0800
From: Jisheng Zhang <Jisheng.Zhang@...aptics.com>
To: Jens Wiklander <jens.wiklander@...aro.org>
Cc: "Steven Rostedt (VMware)" <rostedt@...dmis.org>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
op-tee@...ts.trustedfirmware.org,
Linux ARM <linux-arm-kernel@...ts.infradead.org>
Subject: Re: [PATCH] tee: optee: add invoke_fn tracepoints
On Tue, 23 Feb 2021 08:59:22 +0100 Jens Wiklander wrote:
>
> Hi Jisheng,
Hi Jens,
>
> On Wed, Feb 10, 2021 at 7:44 AM Jisheng Zhang
> <Jisheng.Zhang@...aptics.com> wrote:
> >
> > Add tracepoints to retrieve information about the invoke_fn. This would
> > help to measure how many invoke_fn are triggered and how long it takes
> > to complete one invoke_fn call.
> >
> > Signed-off-by: Jisheng Zhang <Jisheng.Zhang@...aptics.com>
> > ---
> >
> > Since v1:
> > - add BUILD_BUG_ON() macro usage to make sure that the size of what is being
> > copied, is not smaller than the amount being copied. Thank Steve.
> > - move optee_trace.h to keep include headers sorted
> >
> > drivers/tee/optee/call.c | 4 ++
> > drivers/tee/optee/optee_trace.h | 67 +++++++++++++++++++++++++++++++++
> > 2 files changed, 71 insertions(+)
> > create mode 100644 drivers/tee/optee/optee_trace.h
> >
> > diff --git a/drivers/tee/optee/call.c b/drivers/tee/optee/call.c
> > index 780d7c4fd756..0da6fe50f1af 100644
> > --- a/drivers/tee/optee/call.c
> > +++ b/drivers/tee/optee/call.c
> > @@ -14,6 +14,8 @@
> > #include <linux/uaccess.h>
> > #include "optee_private.h"
> > #include "optee_smc.h"
> > +#define CREATE_TRACE_POINTS
> > +#include "optee_trace.h"
> >
> > struct optee_call_waiter {
> > struct list_head list_node;
> > @@ -138,9 +140,11 @@ u32 optee_do_call_with_arg(struct tee_context *ctx, phys_addr_t parg)
> > while (true) {
> > struct arm_smccc_res res;
> >
> > + trace_optee_invoke_fn_begin(¶m);
> > optee->invoke_fn(param.a0, param.a1, param.a2, param.a3,
> > param.a4, param.a5, param.a6, param.a7,
> > &res);
> > + trace_optee_invoke_fn_end(¶m, &res);
> >
> > if (res.a0 == OPTEE_SMC_RETURN_ETHREAD_LIMIT) {
> > /*
> > diff --git a/drivers/tee/optee/optee_trace.h b/drivers/tee/optee/optee_trace.h
> > new file mode 100644
> > index 000000000000..7c954eefa4bf
> > --- /dev/null
> > +++ b/drivers/tee/optee/optee_trace.h
> > @@ -0,0 +1,67 @@
> > +/* SPDX-License-Identifier: GPL-2.0 */
> > +/*
> > + * optee trace points
> > + *
> > + * Copyright (C) 2021 Synaptics Incorporated
> > + * Author: Jisheng Zhang <jszhang@...nel.org>
> > + */
> > +
> > +#undef TRACE_SYSTEM
> > +#define TRACE_SYSTEM optee
> > +
> > +#if !defined(_TRACE_OPTEE_H) || defined(TRACE_HEADER_MULTI_READ)
> > +#define _TRACE_OPTEE_H
> > +
> > +#include <linux/arm-smccc.h>
> > +#include <linux/tracepoint.h>
> > +#include "optee_private.h"
> > +
>
> Checkpatch has some complaints below. Is that something that could be
> fixed or is this so far from regular C-syntax that we don't care?
I tried ./scripts/checkpatch.pl in Linus tree to check the patch, there's
no any error, and except the "MAINTAINERS need updating" warning, there's
no other warnings.
git log -- scripts/checkpatch.pl
shows the latest checkpatch.pl is at commit 62137364e3e8afcc745846c5c67cacf943149073
I'm not sure what happened.
Thanks
>
> Thanks,
> Jens
>
> > +TRACE_EVENT(optee_invoke_fn_begin,
> > + TP_PROTO(struct optee_rpc_param *param),
> > + TP_ARGS(param),
> > +
> > + TP_STRUCT__entry(
> > + __field(void *, param)
> > + __array(u32, args, 8)
> > + ),
> > +
> > + TP_fast_assign(
> > + __entry->param = param;
> > + BUILD_BUG_ON(sizeof(*param) < sizeof(__entry->args));
> > + memcpy(__entry->args, param, sizeof(__entry->args));
> > + ),
> > +
> > + TP_printk("param=%p (%x, %x, %x, %x, %x, %x, %x, %x)", __entry->param,
> > + __entry->args[0], __entry->args[1], __entry->args[2],
> > + __entry->args[3], __entry->args[4], __entry->args[5],
> > + __entry->args[6], __entry->args[7])
> > +);
> > +
> > +TRACE_EVENT(optee_invoke_fn_end,
> > + TP_PROTO(struct optee_rpc_param *param, struct arm_smccc_res *res),
> > + TP_ARGS(param, res),
> > +
> > + TP_STRUCT__entry(
> > + __field(void *, param)
> > + __array(unsigned long, rets, 4)
> > + ),
> > +
> > + TP_fast_assign(
> > + __entry->param = param;
> > + BUILD_BUG_ON(sizeof(*res) < sizeof(__entry->rets));
> > + memcpy(__entry->rets, res, sizeof(__entry->rets));
> > + ),
> > +
> > + TP_printk("param=%p ret (%lx, %lx, %lx, %lx)", __entry->param,
> > + __entry->rets[0], __entry->rets[1], __entry->rets[2],
> > + __entry->rets[3])
> > +);
> > +#endif /* _TRACE_OPTEE_H */
> > +
> > +#undef TRACE_INCLUDE_PATH
> > +#define TRACE_INCLUDE_PATH .
> > +#undef TRACE_INCLUDE_FILE
> > +#define TRACE_INCLUDE_FILE optee_trace
> > +
> > +/* This part must be outside protection */
> > +#include <trace/define_trace.h>
> > --
> > 2.30.0
> >
Powered by blists - more mailing lists