Messages in this thread |  | | Date | Mon, 16 Feb 2015 21:09:11 +0100 | From | Robert Abel <> | Subject | Re: [PATCH 2/4] ARM OMAP2+ GPMC: always program GPMCFCLKDIVIDER |
| |
Hi Tony,
On 16 Feb 2015 18:10, Tony Lindgren wrote: > * Robert ABEL <rabel@cit-ec.uni-bielefeld.de> [150216 07:52]: >> GPMC uses GPMCFCLKDIVIDER during synchronous as well as asynchronous accesses >> in conjunction with WAITMONITORINGTIME. Thus, it's wrong to only program it for >> synchronous accesses. Remove the conditional. > Do you have some test case that gets fixed by this? Or does this > fix some regression? > > The reason why I'm asking is that AFAIK we've had async access > working all along? Maybe to clarify: This is only affects asynchronous accesses with WAITREADMONITORING and/or WAITWRITEMONITORING enabled. Regular asynchronous accesses without WAIT pin monitoring were never affected to begin with and are not affected by this change.
I don't have an off-hand test case and seeing as the current implementation is faulty, it's probable nobody actively uses this feature. Cf. OMAP35xx TRM (SPRUF98X) pp. 1103. Basically, WAITMONITORINGTIME extends the number of cycles from the end of WAIT asserted to when the data is provided/sampled and the read/write access (should) take place. While this behavior is esoteric, it's not a reason not to fix this oversight.
Regards,
Robert
|  |