From 7020d2e35226da7315977b3d78e05edc5da9c6cc Mon Sep 17 00:00:00 2001 From: Matt Mascarenhas Date: Sun, 24 Dec 2017 19:18:01 +0000 Subject: [PATCH] Annotate riscy056 --- miotatsu/riscy/riscy/riscy056.hmml | 73 ++++++++++++++++++++++++++++++ 1 file changed, 73 insertions(+) create mode 100644 miotatsu/riscy/riscy/riscy056.hmml diff --git a/miotatsu/riscy/riscy/riscy056.hmml b/miotatsu/riscy/riscy/riscy056.hmml new file mode 100644 index 0000000..b3584a7 --- /dev/null +++ b/miotatsu/riscy/riscy/riscy056.hmml @@ -0,0 +1,73 @@ +[video member=miotatsu stream_platform=twitch project=riscy title="Trying to Get Our PWM Interrupts to Fire" vod_platform=youtube id=ng-JAY75j88 annotator=Miblo] +[0:08][Recap and set the stage for the day with a working toolchain][:speech] +[2:26][Make and upload pwm_speed_test to the HiFive1] +[4:08][Connect to and reset the HiFive1, and see "Hello, World!"][:run] +[6:22][Check the Freedom E SDK's current init.c file[ref + site=GitHub + page=sifive/freedom-e-sdk/blob/master/bsp/env/freedom-e300-hifive1/init.c + url=https://github.com/sifive/freedom-e-sdk/blob/master/bsp/env/freedom-e300-hifive1/init.c] and the forums[ref + site="SiFive Forums" + page="Does anyone else have problems with output?" + url=https://forums.sifive.com/t/does-anyone-else-have-problems-with-output/446] for information on busy-looping to let the uart settle] +[10:04][Thoughts on dynamically setting how long to wait, based on the CPU frequency][:speech] +[11:41][Build, upload but unfortunately fail to connect to the HiFive1] +[13:42][Try to debug our program with gdb[ref + site="SiFive Developers" + page="HiFive1 Getting Started Guide" + url=https://dev.sifive.com/hifive1/hifive1-getting-started-guide/]] +[16:16][Try to print "Hello World!" in the infinite loop] +[17:07][Connect to the HiFive1 and see garbage being printed][:run] +[17:50][Add "\\r" to the end of the printf] +[18:14][Connect and see it printing "Hello, World!"][:run] +[18:44][Revert to performing the "wfi" in the loop, and printing "Hello World!" before it, build and upload it] +[19:15][Reset the HiFive1 and see no output][:run] +[20:19][Comment out the "wfi" and ensure that the while scope loops infinitely] +[21:57][Connect to the HiFive1 and see our "Hello, World!"][:run] +[22:27][Decide against using the "wfi" instruction] +[23:23][Connect to the HiFive1, see the power lights flickering, and reconnect the power cable][:run] +[24:53][Try to upload and connect, but still see no output][:run] +[25:40][Try a full rebuild and upload] +[26:02][Connect to the HiFive1 and see a "Hello, World!" but no pwm interrupt][:run] +[27:02][Closely read through the code, and compare it with the led_fade.c demo] +[30:03][Configure the PWM in ENALWAYS mode and with PWM_CMP0 = 0[ref + site="SiFive Developers" + page="Freedom E300 Platform Reference Manual" + url=https://www.sifive.com/documentation/freedom-soc/freedom-e300-platform-reference-manual/]][:programming :research] +[36:48][Connect to the HiFive1, see the "Hello, World!" but still no interrupts][:run] +[37:29][#include platform.h and set the pwmscp0ip bit[ref + site="SiFive Developers" + page="Freedom E300 Platform Reference Manual" + url=https://www.sifive.com/documentation/freedom-soc/freedom-e300-platform-reference-manual/]][:programming :research] +[41:54][Connect to the HiFive1 and see no change][:run] +[42:44][Scour the Freedom E SDK demos for interrupts that use the PWM][:research] +[44:27][See how the global_interrupts demo handles the interrupts][:research] +[49:45][Determine that the only difference between global_interrupts and our pwm_speed_test is that we are not setting the priority or the compare registers][:speech] +[50:45][Try to use exactly the same PWM configuration as global_interrupts] +[53:08][Connect to and reset our HiFive1, but still see no PWM interrupt][:run] +[54:24][Try to set the priority and enable timer interrupts] +[56:57][Reset our HiFive1 and see a "trap"][:run] +[58:29][Note that global_interrupts is setting a handler for the timer interrupt and calling set_timer()][:research] +[59:20][Introduce handle_m_time_interrupt()] +[1:04:41][Connect to our HiFive1 and see one instance of "handling timer interrupt" before a "trap", yet no "Hello, World!"][:run] +[1:06:29][Search the demos for MTIP and consult demo_gpio for their timer handling][:research] +[1:07:27][Make handle_m_time_interrupt() disable and reenable the MTIP] +[1:10:05][Connect to our HiFive1 and see the same output][:run] +[1:10:29][Make handle_m_time_interrupt() print "leaving timer interrupt"] +[1:11:03][Connect to our HiFive1 and see the "leaving timer interrupt"] +[1:11:34][Comment out the set_csr(mie, MIP_MTIP) call in main()] +[1:11:53][Connect to our HiFive1 and see that we are still trapping][:run] +[1:12:09][Comment out the following set_csr() calls] +[1:12:40][Connect to our HiFive1 and see the "Hello, World!"][:run] +[1:13:13][Uncomment the set_csr(mstatus, MSTATUS_MIE) call in main()] +[1:13:35][Connect to our HiFive1 and hit the trap][:run] +[1:14:14][Try various combinations of interrupts, :run the program and always hit the trap][:programming] +[1:16:27][Comment out everything except the set_csr() and printf() calls] +[1:17:10][Connect to our HiFive1 and still trap][:run] +[1:17:54][Make and upload demo_gpio] +[1:18:49][Reset our HiFive1 and see everything working as expected][:run] +[1:19:12][Read the documentation for some gotcha related to mstatus[ref + site="RISC-V" + page="Draft Privileged ISA Specification v1.9.1" + url=https://riscv.org/specifications/privileged-isa/]][:research] +[1:21:13][Wrap it up with the determination to continue our investigation next time][:speech] +[/video]