**** BEGIN LOGGING AT Thu Jul 05 03:00:02 2018
Jul 05 06:24:39 <zeekhuge> muneeb17: again, if you are planning to implement it this way, it's better to use RPMsg rather than reinventing the wheel.
Jul 05 06:25:38 <abhishek_[m]> +1
Jul 05 06:25:42 <zeekhuge> And even when this new wheel might not really provide any real benefits over the old one.
Jul 05 06:54:22 <muneeb17> zeekhuge, abhishek_[m], Wormo: Can we finalise on the implementation then, whether to read from the regs or use RPMsg?
Jul 05 06:58:06 <muneeb17> For reading from regs, I think the whole function can be implemented in assembly if we do the pre-processing step to replace the strings with index as suggested by Wormo
Jul 05 06:59:50 <muneeb17> Anyways, we will have to decide about the string implementation
Jul 05 07:52:29 <ds2> rpmsg - might be reuseable for other processor types
Jul 05 08:30:21 <Wormo> I vote against reading from regs as well, so go ahead with RPMsg
Jul 05 16:20:13 <zeekhuge> muneeb17: there ?
Jul 05 16:21:19 <muneeb17> zeekhuge: here. Thanks a lot! for the email
Jul 05 16:22:24 <muneeb17> zeekhuge: For 3.2 though, we could __halt and then re-enable also, right?
Jul 05 16:27:42 <zeekhuge> muneeb17: well, you can even use single step mode for it.
Jul 05 16:28:19 <zeekhuge> It pauses the PRU using control-registers
Jul 05 16:28:34 <zeekhuge> Maybe yeah halt_ is better
Jul 05 16:28:34 <muneeb17> yeah, I was trying to say that only
Jul 05 16:29:01 <zeekhuge> It gets interrupted by an interrupt, right ?
Jul 05 16:29:28 <muneeb17> yeah, the PRU will send the interrupt and halt
Jul 05 16:30:23 <muneeb17> kernel driver can then do the processing and write to the control reg
Jul 05 16:35:53 <muneeb17> zeekhuge: Also about 2, How can the PRU tell us the address of the data, without using rpmsg?
Jul 05 16:39:16 <muneeb17> zeekhuge: I mean the reserved address of the DRAM
Jul 05 16:40:15 <zeekhuge> That can be set in the ELF while compiling.
Jul 05 16:40:29 <muneeb17> could we allocate a reserved section in the elf and modify the linker command file to put that in the DRAM at a decided address?
Jul 05 16:40:56 <zeekhuge> Kind of hard coded
Jul 05 16:41:40 <zeekhuge> I think that can be done as well.
Jul 05 16:43:13 <zeekhuge> muneeb17: look into that .cmd file
Jul 05 16:45:09 <zeekhuge> muneeb17: https://zeekhuge.me/post/ptp_blinky/
Jul 05 16:45:10 <zeekhuge> The linker file
Jul 05 16:50:00 <muneeb17> zeekhuge: I think we can put the address in the SECTIONS part like .mysection > [address], page 1
Jul 05 16:51:09 <zeekhuge> muneeb17: can you please run some initial tests to see if all those things ('pragma', 'ELF parsing' and '.cmd') will work the way we are planning ?
Jul 05 16:51:50 <muneeb17> zeekhuge: Ok, I will do it now only
Jul 05 16:52:58 <muneeb17> zeekhuge: instead of pragma we could use this too - __attribute__((section("name")) to store strings in our elf section, right?
Jul 05 16:54:50 <zeekhuge> I think so. Better test them.
Jul 05 16:55:30 <zeekhuge> But wait
Jul 05 16:56:09 <zeekhuge> muneeb17: we are reserving a section using 'pragma'
Jul 05 16:56:37 <zeekhuge> attribute is used to place data in elf section. Right ?
Jul 05 16:57:29 <muneeb17> Oh...yeah, we will have to use both
Jul 05 16:57:50 <zeekhuge> I mean ..
Jul 05 16:57:50 <zeekhuge> 'pragma' sets a dataRam reserved.
Jul 05 16:57:51 <zeekhuge> _attribute_ puts data into elf section
Jul 05 16:57:57 <zeekhuge> Both are different I think
Jul 05 16:59:52 <muneeb17> zeekhuge: have to go for dinner. will come back after that and test those
Jul 05 17:55:40 <abhishek_[m]> Just sent a mail with some concerns about the latency
Jul 05 18:05:39 <Wormo> I sent an email too
**** ENDING LOGGING AT Fri Jul 06 03:00:01 2018