[Sound-open-firmware] [PATCH] trace: dma: remove disable DMA trace option

[Liam Girdwood]

On Fri, 2018-03-16 at 12:43 +0800, Keyon Jie wrote:
> 
> On 2018年03月16日 12:34, Keyon Jie wrote:
> > 
> > 
> > On 2018年03月16日 11:40, Pan, Xiuli wrote:
> > > 
> > > 
> > > On 3/15/2018 23:09, Liam Girdwood wrote:
> > > > DMA based trace is the standard default trace method. mailbox trace is
> > > > no longer supported.
> > > > 
> > > > Signed-off-by: Liam Girdwood <liam.r.girdwood at linux.intel.com>
> > > > ---
> > > >   configure.ac        | 10 --------
> > > >   src/ipc/dma-copy.c  |  2 --
> > > >   src/ipc/intel-ipc.c |  4 ---
> > > >   src/lib/Makefile.am |  6 +----
> > > >   src/lib/trace.c     | 70 
> > > > +----------------------------------------------------
> > > >   5 files changed, 2 insertions(+), 90 deletions(-)
> > > > 
> > > > diff --git a/configure.ac b/configure.ac
> > > > index bb4d2e9f..9b74c6a9 100644
> > > > --- a/configure.ac
> > > > +++ b/configure.ac
> > > > @@ -241,16 +241,6 @@ case "$with_dsp_core" in
> > > >   esac
> > > > -# dma trace support (Optional), dma trace by default
> > > > -AC_ARG_ENABLE([dma-trace],
> > > > -        AS_HELP_STRING([--disable-dma-trace], [Disabled dma trace 
> > > > and use fallback mailbox trace]))
> > > > -
> > > > -AS_IF([test "x$enable_dma_trace" != "xno"], [
> > > > -    AC_DEFINE([CONFIG_DMA_TRACE], [1], [Configure DMA trace])
> > > > -    ])
> > > > -
> > > > -AM_CONDITIONAL(BUILD_DMA_TRACE,  test "x$enable_dma_trace" != "xno")
> > > > -
> > > >   PLATFORM_BOOT_LDR_LDSCRIPT="boot_ldr.x"
> > > >   AC_SUBST(PLATFORM_BOOT_LDR_LDSCRIPT)
> > > 
> > > Hi Liam,
> > > 
> > > I think we should keep this feature. We may need this feature to debug 
> > > the dma related bugs.
> > > If anything goes wrong with the DMA, or we need to enable DMA on new 
> > > platform, the DMA trace will not work and we may need fall back to 
> > > mail box trace to get the debug info.
> > 
> > Agree to reserve the possibility for mailbox trace, we can set default 
> > to use dma trace, which may be helpful for debugging to special cases, 
> 
> sorry, I mean that the 'mailbox trace' may be helpful for ...
> 

We still have a "mailbox" trace mechanism via trace_err() calls. The error
mailbox is persistent and low volume so it's unlikley to overflow too. This can
be used for enabling DMA trace on future platforms.

Liam