PRUSS MFD staging driver: status update
Hi Greg.
I am interested in continuing the work of Subhasish Ghosh regarding TI PRUSS drivers integration on the kernel side. The last LKML reply I'm aware of is yours, 2 years ago, here:
<https://lkml.org/lkml/2011/7/5/548>, and then there was silence.
Just a quick reminder: PRUSS (Programmable Realtime Unit SubSystem) is a subsystem on various TI DSPs/SoCs that contains two independent RISC cores. Each PRU has its own data/instruction RAM. See
<http://processors.wiki.ti.com/index.php/Programmable_Realtime_Unit_Subsystem>.
My main objective is to eventually have a soft UART (which relies on a PRU) integrated into the mainline. Since this would depend on common routines shared by other drivers using PRUSS (enable/disable the PRUSS clock, load a firmware, start/stop a specific PRU, configure interrupts, etc.), common exported PRUSS functions need to be available.
I believe that's what Subhasish Ghosh was trying to do with his PRUSS MFD staging driver.
Now, what is the status of this? Would you still accept a valid patch for this "MFD" driver in staging? My idea, however, is more about a set of common functions that could be used by other "real" drivers. This means nothing exported to user space, no sysfs, etc., since only kernel drivers would use them (I don't intend user space drivers to rely on this; they already have drivers/uio/uio_pruss for this).
Also, at least today, PRUSS is not only part of the DA850 devices, but also included into OMAP-L1x8/OMAP-L1x7, C674n, AM18xx/AM17XX and AM335x devices, so I wouldn't mention "DA8XX" in the Kconfig.
Please Cc me in any answer/comment regarding this message.
Thank you if you have any time to look into this,
Phil
Quoted 6 lines · Philippe Proulx
Hi Greg.
I am interested in continuing the work of Subhasish Ghosh regarding TI PRUSS drivers integration on the kernel side. The last LKML reply I'm aware of is yours, 2 years ago, here:
<https://lkml.org/lkml/2011/7/5/548>, and then there was silence.
Then please follow what I suggested there and resend the code, or send whatever you are proposing. It's hard to judge vague statements without seeing the code as to what you wish to have integrated, don't you agree?
thanks,
greg k-h