LeafLabs Garden » Topic: BIG BUG on interrupt functionality :(

LeafLabs Garden » Topic: BIG BUG on interrupt functionality :( http://forums.leaflabs.com/topic.php?id=714 A place to share, learn, and grow... en-US Fri, 22 Jan 2016 00:17:20 +0000 http://bbpress.org/?v=1.0.2 <![CDATA[Search]]> q http://forums.leaflabs.com/search.php mbolivar on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4146 Mon, 04 Apr 2011 16:23:27 +0000 mbolivar 4146@http://forums.leaflabs.com/ Roberto, <blockquote> @mbolivar What do you think about this bug ? </blockquote> In short, "What snigelen said". The EXTI lines mux together the corresponding GPIO bits, so you can't use PD14 and PE14 to serve different interrupts at the same time. Are you still having any other problems with external interrupts using the latest refactor code? I found and fixed a few bugs while I was redoing that interface, so things should work better now. redfox74 on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4130 Mon, 04 Apr 2011 10:47:27 +0000 redfox74 4130@http://forums.leaflabs.com/ mhh I try to deactivate one of this two pin but the other don't start to work .. now i check again :) Best Roberto snigelen on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4129 Mon, 04 Apr 2011 10:05:57 +0000 snigelen 4129@http://forums.leaflabs.com/ Yes, exactly. And that's why they say this on the LeafLabs EXTI-documentation: <blockquote>In effect, this means that every pin on the Maple maps to a certain EXTI line, and within that EXTI line, only one of the pins that maps to it can be used as an external interrupt at a time.</blockquote> (my emphasis). redfox74 on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4128 Mon, 04 Apr 2011 09:17:02 +0000 redfox74 4128@http://forums.leaflabs.com/ Hi Snigelen do you mean that if i put all bit to 1111 the micro don't reply to all the PIN ? PA14,PB14,PC14,PD14,PE14,PF14,PG14 ? Best Roberto snigelen on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4127 Mon, 04 Apr 2011 08:31:13 +0000 snigelen 4127@http://forums.leaflabs.com/ I think anton is right, each ext.int.line can only be connected to a given pin-number on _one_ port. You say "The right configuration will be : EXTI14[3:0] 0111", but that isn't a valid number in that register (for the Maple, it may be valid on a larger device with more ports). According to the reference manual it's (i replaced x with 14 here): <pre><code>EXTI14[3:0]: EXTI 14 configuration These bits are written by software to select the source input for EXTI14 external interrupt. 0000: PA[14] pin 0001: PB[14] pin 0010: PC[14] pin 0011: PD[14] pin 0100: PE[14] pin 0101: PF[14] pin 0110: PG[14] pin</code></pre> so the bit's isn't a bit mask, it's a port number. So you can only select one of them. redfox74 on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4126 Mon, 04 Apr 2011 08:08:30 +0000 redfox74 4126@http://forums.leaflabs.com/ @mbolivar What do you think about this bug ? Best Roberto redfox74 on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4125 Mon, 04 Apr 2011 08:06:14 +0000 redfox74 4125@http://forums.leaflabs.com/ Hi Anton, why the problem can't be solved ? The document of leaflab was Maple V5 that is medium density , I using High Density , native code ... in the doc The right configuration will be : EXTI14[3:0] 0111 ? Is possible to activate all 2 pin togheter ? What do yo think ? Best Roberto anton on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4124 Mon, 04 Apr 2011 05:34:30 +0000 anton 4124@http://forums.leaflabs.com/ PD14/PE14 problem can't be solved. <a href="http://leaflabs.com/docs/external-interrupts.html" rel="nofollow">http://leaflabs.com/docs/external-interrupts.html</a> <a href="http://www.st.com/stonline/products/literature/rm/13902.pdf" rel="nofollow">http://www.st.com/stonline/products/literature/rm/13902.pdf</a> page 192. Nibble of AFIO_EXTICRx corresponding to pin number should be cleared before OR with new value. pin = 14; port = 3; // PD14 __writexor(afio_regs[pin/4], (port << ((pin % 4) * 4))); writes 0x0300 to AFIO_EXTICR4 next after pin = 14; port = 4; // PE14 __writexor(afio_regs[pin/4], (port << ((pin % 4) * 4))); AFIO_EXTICR4 contents invalid value 0x0700 instead of 0x0400 redfox74 on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4123 Mon, 04 Apr 2011 05:11:02 +0000 redfox74 4123@http://forums.leaflabs.com/ I Try your patch , but don't work and don't solve the problem of PD14 and PE14 afio_regs[pin/4] &= ~(0xF << ((pin % 4) * 4)); afio_regs[pin/4] |= (port << ((pin % 4) * 4)); This is my code #define __write(reg, value) (*(volatile uint32*)(reg) = (value)) // old write #define __writexor(reg, value) (*(volatile uint32*)(reg) |= (value)) // write with or in register. Best Roberto anton on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4121 Sun, 03 Apr 2011 22:25:23 +0000 anton 4121@http://forums.leaflabs.com/ If you call attach_interript with |= for PD14 and PE14 it will connect 14 exti chanel to missing PH14 pin. And this bug is patched in the refactor branch: <pre><code>void afio_exti_select(afio_exti_num exti, afio_exti_port gpio_port) { __io uint32 *exti_cr = &AFIO_BASE->EXTICR1 + exti / 4; uint32 shift = 4 * (exti % 4); uint32 cr = *exti_cr; cr &= ~(AFIO_EXTI_SEL_MASK << shift); cr |= gpio_port << shift; *exti_cr = cr; }</code></pre> anton on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4120 Sun, 03 Apr 2011 20:13:23 +0000 anton 4120@http://forums.leaflabs.com/ Yes, there is 16 EXTI cannels and chanel number corresponds pin number. attach_interript with the same pin number rewrites AFIO_EXTICRx. I'll try your patch. I think it should look like <pre><code>afio_regs[pin/4] &= ~(0xF << ((pin % 4) * 4)); afio_regs[pin/4] |= (port << ((pin % 4) * 4));</code></pre> redfox74 on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4119 Sun, 03 Apr 2011 19:01:25 +0000 redfox74 4119@http://forums.leaflabs.com/ I doing the test .. So it's better of my last test , now there're only 2 pin that don' work under interrupt pin 89 and pin 57 that correspond at : PD14 and PE14 ... mhhh the two pin are 14 ... ?!? Could be this the answer at my problem . Tomorrow will check ... my 2 cents to this good project ;) Best Roberto redfox74 on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4118 Sun, 03 Apr 2011 18:02:43 +0000 redfox74 4118@http://forums.leaflabs.com/ Dear Friends, I found the problem in __write function , it update the register only with the last pin and don't doing an |= xor with other pin yet set. This is the patch ... uint32 channel = pin; /* map port to channel */ // __write(afio_regs[pin/4], (port << ((pin % 4) * 4))); __writexor(afio_regs[pin/4], (port << ((pin % 4) * 4))); // i add in util.h the function #define __writexor(reg, value) (*(volatile uint32*)(reg) |= (value)) add an or ... so is better of first test .. but i think that there are other problem again ... /* Unmask appropriate interrupt line */ __set_bits(EXTI_IMR, BIT(channel)); redfox74 on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4116 Sun, 03 Apr 2011 07:13:18 +0000 redfox74 4116@http://forums.leaflabs.com/ Can other people doing the test on Maple V5 ? Best Roberto redfox74 on "BIG BUG on interrupt functionality :(" http://forums.leaflabs.com/topic.php?id=714#post-4114 Sat, 02 Apr 2011 19:24:49 +0000 redfox74 4114@http://forums.leaflabs.com/ So i testing a functionality for manage a multiple interrupt on more port ... the code is simple : #define CH1 22 #define CH2 23 #define CH3 24 #define CH4 89 #define CH5 59 #define CH6 62 #define CH7 57 #define CH8 60 void rx1(void) { if (digitalRead(CH1)) { Serial4.print("1UP"); } else { Serial4.print("1DOWN"); } } void rx2(void) { if (digitalRead(CH2)) { Serial4.print("2UP"); } else { Serial4.print("2DOWN"); } } void rx3(void) { if (digitalRead(CH3)) { Serial4.print("3UP"); } else { Serial4.print("3DOWN"); } } void rx4(void) { if (digitalRead(CH4)) { Serial4.print("4UP"); } else { Serial4.print("4DOWN"); } } void rx5(void) { if (digitalRead(CH5)) { Serial4.print("5UP"); } else { Serial4.print("5DOWN"); } } void rx6(void) { if (digitalRead(CH6)) { Serial4.print("6UP"); } else { Serial4.print("7DOWN"); } } void rx7(void) { if (digitalRead(CH7)) { Serial4.print("7UP"); } else { Serial4.print("7DOWN"); } } pinMode(CH1,INPUT); pinMode(CH2,INPUT); pinMode(CH3,INPUT); pinMode(CH4,INPUT); pinMode(CH5,INPUT); pinMode(CH6,INPUT); pinMode(CH7,INPUT); pinMode(CH8,INPUT); attachInterrupt(CH1, rx1, RISING); attachInterrupt(CH2, rx2, RISING); attachInterrupt(CH3, rx3, RISING); attachInterrupt(CH4, rx4, RISING); attachInterrupt(CH5, rx5, RISING); attachInterrupt(CH6, rx6, RISING); attachInterrupt(CH7, rx7, RISING); If i activate all interrupt ... only CH1 CH2 CH3 CH7 work fine , the other interrupt don't work ... channel CH4 CH5 Ch6 not working properly. Instead if i activate only one channel and test all pin the interrupt function work fine ... when i testing all interrupt i check only 1 input port then the other input port ecc don't all port togheter .. So Where is the problem ? My idea is that there re some problem on exti.c in this portion of code : void __irq_exti9_5(void) { /* Figure out which channel it came from */ uint32 pending; uint32 i; pending = REG_GET(EXTI_PR); pending = GET_BITS(pending, 5, 9); /* Dispatch every handler if the pending bit is set */ for (i = 0; i < 5; i++) { if (pending & 0x1) { dispatch_handler(EXTI5 + i); clear_pending(EXTI5 + i); } pending >>= 1; } } void __irq_exti15_10(void) { /* Figure out which channel it came from */ uint32 pending; uint32 i; pending = REG_GET(EXTI_PR); pending = GET_BITS(pending, 10, 15); /* Dispatch every handler if the pending bit is set */ for (i = 0; i < 6; i++) { if (pending & 0x1) { dispatch_handler(EXTI10 + i); clear_pending(EXTI10 + i); } pending >>= 1; } } Best Roberto