For the last two months I'm trying to understand the logic behind ARM MCUs, so I ran across many tutorials. By far you have the best presentations! Thank you my friend!
Although the voice in the video is not perfect, it's surprisingly good for a robot, and considering the content and explaination was excellent, I'm very happy with the final product! Subscribed.
@16:18 so the process of preserving environment variable(by pushing the aforementioned registers onto the stack) cost the CPU 32-bits of addresses.. while unstacking, does it cost cpu 32-bits of address line too? And also does ISR only do two instructions? (BX and LR).. How much address the CPU need for this?.. So why does we need to offset 64bits of address when we want to determine the address of program ISR? (64+4xn) From what I understand, stacking environment variables cost 32bits of address, which corresponds to 4 bits of address is used to push a value from a register to the stack.. Does this means BX and LX cost the CPU 4 bits of address each? If that the case to find the next address of ISR program would be 32+4+32+4xn. if not then, 32(for stacking)+32(for unstacking)+4(for BX and LX or any other ISR operations that might require only 4bits of address)*n(nth element of an array of ISR)
@15:41 why he is saying that the program ISR (Interrupt Service Routine) or Interrupt Handler has the size of 32 bytes but demonstrate it in 32 bits? if 1 opcode is one byte(8bits), and 1 data line is 64 bits.. 1 bit of address line can have an instruction(opcode) + data =64 bits 4 bits of address line= 4*64 =256 bits or 1 byte.. 8 bits of address line= 2 bytes if 32 bits of address line= 8 bytes.. oh ok so i understand this ISR program requires 8 bytes of address line.. Or 32 address space
Thanks for the nice video. In the nested interrupt example, why isn't Pending Register for EXTI3 becomes 1 again when it is stacked and waited for DMA1_Channel2 interrupts to be finished?
13:46 Isn't Stack Last-in-First-out? According to explanation here, it looks like registers are stacked inversely because top of the stack is popped first.
in 17:21 there shuldn't be a pending state. I observed in my program. If you are pointing that this pending bit is flipping 1 to 0 so quickly when we enter ISR12 that we cannot observe this case, again there is no 1 to 0 switch. Could you provide me your proof?
This is an interesting question. The "Interrupt Set-pending Registers" and "Interrupt Clear-pending Registers" works in pairs to set or clear the pending state of an interrupt. You can find more information in Cortex -M4 Devices Generic User Guide.
@@embeddedsystemswitharmcort9051 sorry for my bad english but did you completelly understand what i am tying to tell you? If i couldnt, i can tell with other sentence. I appriceate your videos so i respect you a lot. Could you provide me the exact sentence in the document?
15:38 When an interrupt is done, a BX LR instruction is executed? But how does LR register receives the value of PC before the interrupt? Does it happen automatically? Or we need to get it from the stack thst was previously saved?
@@embeddedsystemswitharmcort9051 Thank you for the fast answer. I think saying that exceptional return value EXC_RETURN is stored there during the interrupt handling is better. Loading this value by BX will automatically start unstacking and subsequently load the right value to PC. I've done some research on my own :)
There is a key piece of information here which the ARM documentation on the NVIC does NOT answer: Is a higher priority number more or less urgent? Thank you for including the answer in this video!
Lower value means higher priority. Configurable values range from 0 to 15, while non-configurable high-priority interrupts can even have negative values. -3 being the highest priority. You can check these informations in the RM0090 reference manual under "table 61." ("External interrupt/event controller (EXTI)" section).
The slide is correct. The address to jump to must have the least significant bit of the least significant bite set to 1 to indicate the processor is executing in THUMB mode. The video does explain this.
17:30 -- Is it possible to disable Interrupt nesting in the ARM Cortex-M Microcontrollers? Is it required to be performed in software? Or is there a way to configure the NVIC to be able to block the other IRQs once a IRQ is being serviced?
For the last two months I'm trying to understand the logic behind ARM MCUs, so I ran across many tutorials. By far you have the best presentations!
Thank you my friend!
Thank you robot for the clear explanation, using an almost dead, emotionless voice.
Although the voice in the video is not perfect, it's surprisingly good for a robot, and considering the content and explaination was excellent, I'm very happy with the final product! Subscribed.
Glad to know that. I have started to use my own voice now.
This is a pure gold .
Excellent Explanation, Thank you very much Prof Yifeng
Thanks! Very well explained!
Voice like a melted butter
Yeah, in the future, I will use my own voice.
@16:18
so the process of preserving environment variable(by pushing the aforementioned registers onto the stack)
cost the CPU 32-bits of addresses..
while unstacking, does it cost cpu 32-bits of address line too?
And also does ISR only do two instructions? (BX and LR)..
How much address the CPU need for this?..
So why does we need to offset 64bits of address when we want to determine the address of program ISR? (64+4xn)
From what I understand, stacking environment variables cost 32bits of
address, which corresponds to 4 bits of address is used to push a value from a register to the stack..
Does this means BX and LX
cost the CPU 4 bits of address each?
If that the case to find the next address of ISR program would be
32+4+32+4xn.
if not then,
32(for stacking)+32(for unstacking)+4(for BX and LX or any other ISR operations that might require only 4bits of address)*n(nth element of an array of ISR)
In the interrupt handler function why are we setting the Bit in PR register at last ?
@15:41
why he is saying that the program ISR (Interrupt Service Routine) or Interrupt Handler has the size of 32 bytes but demonstrate it in 32 bits?
if 1 opcode is one byte(8bits), and 1 data line is 64 bits..
1 bit of address line can have an instruction(opcode) + data =64 bits
4 bits of address line= 4*64 =256 bits or 1 byte..
8 bits of address line= 2 bytes
if 32 bits of address line= 8 bytes..
oh ok so i understand this ISR program requires 8 bytes of address line..
Or 32 address space
10:48 why in the address 0x00000064 does it contain the address 0x0800030D and not 0x0800030C ?
can anyone help please
The least significant bit must be 1 to indicate that the processor is to run in Thumb mode. ARM Cortex-M only support Thumb instructions.
Excellent Presentation about NVIC i have ever seen...............................
64 + 4 *n , could you clarify the formula in more details, please? Thanks a million.
Thank you! very clear and helpfull!
What a great presentation! Excelent work!
Thank you. I am glad you like it.
can we get the slides please? thank you for the video.
9:10 - Interrupt Vector Table & Example
this channel is underrated
Thank you Dr. Zhu
when dma interrupt is performing whether exti3 is in active state or pending state
What robot software reads that text? It has an amazing natural pronunciation
Wow. Now I understand interrupts for my job interview XD.
did you get the job?
@@da_druuskee7709 yes I got it. :)
i'm reviewing for that right now! Any advice? XD
are you on LinkedIn. i sent you a request please accept i am From India
Man, I owe you a beer. Really.
Buy his book then
@@minghanqiu1095 I did!
you make the ARM architecture story sound so clear
Very good analogy for polling vs interrupt.
Thanks for the nice video. In the nested interrupt example, why isn't Pending Register for EXTI3 becomes 1 again when it is stacked and waited for DMA1_Channel2 interrupts to be finished?
Very good class with several amazing details !!!
at 1:23 the phone analogy doesn't really work. If you pick up the phone before a call arrives then the sender will get a busy tone.
You pick up the phone and hang it up if there is no incoming call.
Wahuuuu amazing video Prof. Zhu... Thanks a lot..!
Most welcome!
Outstanding video, thank you very much
13:46 Isn't Stack Last-in-First-out? According to explanation here, it looks like registers are stacked inversely because top of the stack is popped first.
Cortex-M4 uses full descending stack by default.
@@embeddedsystemswitharmcort9051 yes that is true. You are correct. I had lack of conceptual knowledge
in 17:21 there shuldn't be a pending state. I observed in my program. If you are pointing that this pending bit is flipping 1 to 0 so quickly when we enter ISR12 that we cannot observe this case, again there is no 1 to 0 switch. Could you provide me your proof?
This is an interesting question. The "Interrupt Set-pending Registers" and "Interrupt Clear-pending Registers" works in pairs to set or clear the pending state of an interrupt. You can find more information in Cortex -M4 Devices Generic User Guide.
@@embeddedsystemswitharmcort9051 sorry for my bad english but did you completelly understand what i am tying to tell you? If i couldnt, i can tell with other sentence. I appriceate your videos so i respect you a lot. Could you provide me the exact sentence in the document?
in intrupt calculation
what is mean by 64
15:38 When an interrupt is done, a BX LR instruction is executed? But how does LR register receives the value of PC before the interrupt? Does it happen automatically? Or we need to get it from the stack thst was previously saved?
The hardware automatically preserves LR value onto the stack, and sets LR to a new value to indicate which stack is used. This happens automatically.
@@embeddedsystemswitharmcort9051 Thank you for the fast answer. I think saying that exceptional return value EXC_RETURN is stored there during the interrupt handling is better. Loading this value by BX will automatically start unstacking and subsequently load the right value to PC. I've done some research on my own :)
There is a key piece of information here which the ARM documentation on the NVIC does NOT answer: Is a higher priority number more or less urgent? Thank you for including the answer in this video!
Great to know that.
Lower value means higher priority. Configurable values range from 0 to 15, while non-configurable high-priority interrupts can even have negative values. -3 being the highest priority. You can check these informations in the RM0090 reference manual under "table 61." ("External interrupt/event controller (EXTI)" section).
Great video
Does NVIC clear the icpr register by itself? Or is it the responsability of the interrupt handler?
Great explanaition. Thank you for your effort and great teaching skills. Best wishes.
These are great videos. Is the book having similar explanation?
A very good lesson .... you explain so well ... thanks and GOD bless.
perfect movie for a rainy day
yes, with a girlfried in the arms, on the sofa and a cup of potatochips, fine selection of cheese and a bottle Chateau Lafite 1936...
@@paulg.3067
incredible explanation
wait is there an error at 11:42 because the address of the pointer to the ISR is 30D but the address of the ISR is 30C? What is going on here?
The slide is correct. The address to jump to must have the least significant bit of the least significant bite set to 1 to indicate the processor is executing in THUMB mode. The video does explain this.
Thanks sir for making this video. keep uploading more.
17:30 -- Is it possible to disable Interrupt nesting in the ARM Cortex-M Microcontrollers? Is it required to be performed in software? Or is there a way to configure the NVIC to be able to block the other IRQs once a IRQ is being serviced?
There is never a reason for this as long as you sort priorities according to needs.
Priority vs. subpriority depth is a compromise at cost of subpriority bit depth, and is configurable in most Cortex-Ms.
So what is preemption
In explain sram mapping, it states stack going down, heap up. That,s only one stack. When in MSP and PSP, how sram is configured?
MSP and PSP are configured by software during the booting process, specifically by startup.s.
Ah; $67 without shipping for the book. You should publish an epub version Pr. Shu
very good tutorial!
Only complain with you is Robo voice, please upload same video in your original voice
Can you explan that why stacking and unstacking take 12 cycle , again tail chain take 6 cycle? Please ...
community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/beginner-guide-on-interrupt-latency-and-interrupt-latency-of-the-arm-cortex-m-processors
thank you very much
thank you for such great explanation.
You are welcome!
Awesome video thank you
Excellent Explanation, Thank you very much Prof Yifeng~~~
Can I Get The PPT?
very good tutorial video!!!
awesome lecture! thanks!
You are welcome!
Please consider narrating the videos instead of using an automatic voice. Other than that, great video!
Noted!
Is a robot reading the text?
yes, its to eliminate people potentially complaining about Prof Yifeng Zhu's accent
Robots everywhere these days....
@@superiorphi5778 his accent isnt even that bad ... :/
better robot than strong accent
RUclips does not allow voice over an existing video. Do you folks know a good solution?
what is meant by 64 in interrupt calculation (64+4*n)?
That Cortex M-4 system must be big-endian, no doubt!
excellent
Watch the video at 1.5 or 2.0 speed.
i emailed youtube to add 3 and 3.5 speeds
thank u
Sciencephille the AI? what are you doing here?
The automatic voice is unbearable
please replace the audio
hi bro What is the maximum interrupt speed for stm32f767zi?
Can't listen to this synthesized voice...
downvote for robot voice
what was he saying??