|
Hi, We are having problems to download the .out when using XDS560 emulator. We have a C6203 board and when we try to download the .out to the DSP via XDS560, a message appears warning that the endianess of the target is different from the endianess of the .out. The DSP was on-board programmed, via pull-up/pull-down, to work in little-endian, but when we read CSR register, via Code Composer, it's 8th bit is in '0', indicating big-endian. The DSP pin that programs the endianess is connected to a PLX PCI bridge pin, that drives the bus after the DSP reset. The DSP reads the pin value during its reset, so in normal functioning (without emulator), the endianess is correctly read by the DSP. But when we use the emulator, it seems that XDS560 reads the pin value to determine the endianess, but at this time the bus belongs to the PLX, that probably drives the pin to '0'. I think that the correct would be if XDS560 read the CSR value, not the pin value. Anyone could help me to solve this problem, i.e. what can I do in order to force the XDS560 to read correctly the endianess? Thank you, Regards, Alexandre |
|
|
Problem loading code with XDS560
Started by ●June 23, 2003
Reply by ●June 23, 20032003-06-23
|
Hi, Jeff, How can I do this? Is CSR memory mapped? Alexandre -----Original Message----- From: Jeff Brower [mailto:] Sent: Monday, June 23, 2003 1:24 PM To: Alexandre Freire da Silva Osorio Cc: Subject: Re: [c6x] Problem loading code with XDS560 Alexandre Freire da Silva Osorio- > We are having problems to download the .out when using XDS560 emulator. > > We have a C6203 board and when we try to download the .out to the DSP > via XDS560, a message appears warning that the endianess of the target > is different from the endianess of the .out. The DSP was on-board > programmed, via pull-up/pull-down, to work in little-endian, but when > we read CSR register, via Code Composer, it's 8th bit is in '0', > indicating big-endian. > > The DSP pin that programs the endianess is connected to a PLX PCI > bridge pin, that drives the bus after the DSP reset. The DSP reads > the pin value during its reset, so in normal functioning (without > emulator), the endianess is correctly read by the DSP. But when we > use the emulator, it seems that XDS560 reads the pin value to > determine the endianess, but at this time the bus belongs to the PLX, > that probably drives the pin to '0'. I think that the correct would > be if XDS560 read the CSR value, not the pin value. > > Anyone could help me to solve this problem, i.e. what can I do in > order to force the XDS560 to read correctly the endianess? Can you set the CSR register in a GEL file, prior to downloading code? Jeff Brower system engineer Signalogic |
Reply by ●June 23, 20032003-06-23
|
Alexandre Freire da Silva Osorio- > We are having problems to download the .out when using XDS560 emulator. > > We have a C6203 board and when we try to download the .out to the DSP > via XDS560, a message appears warning that the endianess of the target > is different from the endianess of the .out. The DSP was on-board > programmed, via pull-up/pull-down, to work in little-endian, but when > we read CSR register, via Code Composer, it's 8th bit is in '0', > indicating big-endian. > > The DSP pin that programs the endianess is connected to a PLX PCI > bridge pin, that drives the bus after the DSP reset. The DSP reads > the pin value during its reset, so in normal functioning (without > emulator), the endianess is correctly read by the DSP. But when we > use the emulator, it seems that XDS560 reads the pin value to > determine the endianess, but at this time the bus belongs to the PLX, > that probably drives the pin to '0'. I think that the correct would > be if XDS560 read the CSR value, not the pin value. > > Anyone could help me to solve this problem, i.e. what can I do in > order to force the XDS560 to read correctly the endianess? Can you set the CSR register in a GEL file, prior to downloading code? Jeff Brower system engineer Signalogic |
Reply by ●June 24, 20032003-06-24
|
Hi, Alexandre Freire da Silva Osorio, The key problem is the time order of reset of dsp and the state of the endianess configuration pin. The pin must be high if you need little endian during the reset of dsp. Each time you reset(manual, power on) you dsp the pin is read and write to CSR. So why don't you directly pull it up? The endianess bit in CSR is read only. So it is not problem of software, it is your hardware. Have a scope to check the state of the configuration pin while the dsp is in reset. Wang Tao ----- Original Message ----- From: "Alexandre Freire da Silva Osorio" <> Date: Mon, 23 Jun 2003 12:02:20 -0300 To: <> Subject: [c6x] Problem loading code with XDS560 Re: Hi, Re: Re: We are having problems to download the .out when using XDS560 emulator. Re: Re: We have a C6203 board and when we try to download the .out to the DSP via XDS560, a message appears warning that the endianess of the target is different from the endianess of the .out. The DSP was on-board programmed, via pull-up/pull-down, to work in little-endian, but when we read CSR register, via Code Composer, it's 8th bit is in '0', indicating big-endian. Re: Re: The DSP pin that programs the endianess is connected to a PLX PCI bridge pin, that drives the bus after the DSP reset. The DSP reads the pin value during its reset, so in normal functioning (without emulator), the endianess is correctly read by the DSP. But when we use the emulator, it seems that XDS560 reads the pin value to determine the endianess, but at this time the bus belongs to the PLX, that probably drives the pin to '0'. I think that the correct would be if XDS560 read the CSR value, not the pin value. Re: Re: Anyone could help me to solve this problem, i.e. what can I do in order to force the XDS560 to read correctly the endianess? Re: Re: Thank you, Re: Re: Regards, Re: Alexandre Re: Re: Re: Re: _____________________________________ Re: Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. Re: Re: _____________________________________ Re: About this discussion group: Re: Re: To Join: Send an email to Re: Re: To Post: Send an email to Re: Re: To Leave: Send an email to Re: Re: Archives: http://www.yahoogroups.com/group/c6x Re: Re: Other Groups: http://www.dsprelated.com Re: Re: Re: ">http://docs.yahoo.com/info/terms/ Re: Re: -- __________________________________________________________ Sign-up for your own FREE Personalized E-mail at Mail.com http://www.mail.com/?sr=signup CareerBuilder.com has over 400,000 jobs. Be smarter about your job search http://corp.mail.com/careers |
Reply by ●June 24, 20032003-06-24
|
Hi all, Tao has given the right answer. When the pin is "pulled up", the board will be in little endian and when it is down it will be in big endian. CSR can't be set by software as it is a read only bit. The 8th bit as Alexandre pointed out denotes the endianness. Hence when the power is switched on to the board, the status of the pin determines your endianness. BTW I have a doubt. Are you trying a little endian program or a big endian program. I work on C6416 with same configuration but haven't had success with big endian board and XDS560. There seems to be some problem with XDS560 drivers. Actually for little endian mode, it works without a glitch but for big endian mode, it opens the CCS but when I try to load the program, it gives a problem with respect to memory mapping. Actually the GEL file provided with the XDS560 driver doesn't initialize the memories or rather memory-mapping doesn't occur. when I modify the same GEL file with all the memory mapping initializations, it doesn't open the CCS at all and hangs the CCS. I would like to know if anyone else has succeeded in loading and running a big endian program on the board with XDS560. If so, can you share the GEL file with me. Also hope this clarifies your doubt. Thanks and Regards, Ganesh ----- Original Message ----- From: Tao Wang To: Alexandre Freire da Silva Osorio ; Sent: Tuesday, June 24, 2003 6:49 AM Subject: Re: [c6x] Problem loading code with XDS560 Hi, Alexandre Freire da Silva Osorio, The key problem is the time order of reset of dsp and the state of the endianess configuration pin. The pin must be high if you need little endian during the reset of dsp. Each time you reset(manual, power on) you dsp the pin is read and write to CSR. So why don't you directly pull it up? The endianess bit in CSR is read only. So it is not problem of software, it is your hardware. Have a scope to check the state of the configuration pin while the dsp is in reset. Wang Tao ----- Original Message ----- From: "Alexandre Freire da Silva Osorio" <> Date: Mon, 23 Jun 2003 12:02:20 -0300 To: <> Subject: [c6x] Problem loading code with XDS560 Re: Hi, Re: Re: We are having problems to download the .out when using XDS560 emulator. Re: Re: We have a C6203 board and when we try to download the .out to the DSP via XDS560, a message appears warning that the endianess of the target is different from the endianess of the .out. The DSP was on-board programmed, via pull-up/pull-down, to work in little-endian, but when we read CSR register, via Code Composer, it's 8th bit is in '0', indicating big-endian. Re: Re: The DSP pin that programs the endianess is connected to a PLX PCI bridge pin, that drives the bus after the DSP reset. The DSP reads the pin value during its reset, so in normal functioning (without emulator), the endianess is correctly read by the DSP. But when we use the emulator, it seems that XDS560 reads the pin value to determine the endianess, but at this time the bus belongs to the PLX, that probably drives the pin to '0'. I think that the correct would be if XDS560 read the CSR value, not the pin value. Re: Re: Anyone could help me to solve this problem, i.e. what can I do in order to force the XDS560 to read correctly the endianess? Re: Re: Thank you, Re: Re: Regards, Re: Alexandre Re: Re: Re: Re: _____________________________________ Re: Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. Re: Re: _____________________________________ Re: About this discussion group: Re: Re: To Join: Send an email to Re: Re: To Post: Send an email to Re: Re: To Leave: Send an email to Re: Re: Archives: http://www.yahoogroups.com/group/c6x Re: Re: Other Groups: http://www.dsprelated.com Re: Re: Re: ">http://docs.yahoo.com/info/terms/ Re: Re: -- __________________________________________________________ Sign-up for your own FREE Personalized E-mail at Mail.com http://www.mail.com/?sr=signup CareerBuilder.com has over 400,000 jobs. Be smarter about your job search http://corp.mail.com/careers _____________________________________ Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. _____________________________________ About this discussion group: To Join: Send an email to To Post: Send an email to To Leave: Send an email to Archives: http://www.yahoogroups.com/group/c6x Other Groups: http://www.dsprelated.com ">http://docs.yahoo.com/info/terms/ |
Reply by ●June 24, 20032003-06-24
|
Ganesh and Tao, Thank you very much for your answers. The endianess pin in our board is already pulled-up (little-endian). So, the problem seems to be the XDS560. It should read the 8th CSR bit, not the pin directly. When we power-up the board, the DSP reads the pin as logical "1", because PLX is not the bus master at this time. After this, we cannot guarantee the pin value, because it is being mastered by PLX. In respect to download a big-endian program, yes, we tried it and had the same problem as Ganesh did. Many memory locations was not correctly loaded. We tried it after we concluded that XDS would not correctly read the DSP endianess. Thank you. regards, Alexandre -----Original Message----- From: Ganesh Vijayan [mailto:] Sent: Tuesday, June 24, 2003 1:46 AM To: Alexandre Freire da Silva Osorio; ; Tao Wang Subject: Re: [c6x] Problem loading code with XDS560 Hi all, Tao has given the right answer. When the pin is "pulled up", the board will be in little endian and when it is down it will be in big endian. CSR can't be set by software as it is a read only bit. The 8th bit as Alexandre pointed out denotes the endianness. Hence when the power is switched on to the board, the status of the pin determines your endianness. BTW I have a doubt. Are you trying a little endian program or a big endian program. I work on C6416 with same configuration but haven't had success with big endian board and XDS560. There seems to be some problem with XDS560 drivers. Actually for little endian mode, it works without a glitch but for big endian mode, it opens the CCS but when I try to load the program, it gives a problem with respect to memory mapping. Actually the GEL file provided with the XDS560 driver doesn't initialize the memories or rather memory-mapping doesn't occur. when I modify the same GEL file with all the memory mapping initializations, it doesn't open the CCS at all and hangs the CCS. I would like to know if anyone else has succeeded in loading and running a big endian program on the board with XDS560. If so, can you share the GEL file with me. Also hope this clarifies your doubt. Thanks and Regards, Ganesh ----- Original Message ----- From: Tao Wang <mailto:> To: Alexandre Freire da Silva Osorio <mailto:> ; Sent: Tuesday, June 24, 2003 6:49 AM Subject: Re: [c6x] Problem loading code with XDS560 Hi, Alexandre Freire da Silva Osorio, The key problem is the time order of reset of dsp and the state of the endianess configuration pin. The pin must be high if you need little endian during the reset of dsp. Each time you reset(manual, power on) you dsp the pin is read and write to CSR. So why don't you directly pull it up? The endianess bit in CSR is read only. So it is not problem of software, it is your hardware. Have a scope to check the state of the configuration pin while the dsp is in reset. Wang Tao ----- Original Message ----- From: "Alexandre Freire da Silva Osorio" < > Date: Mon, 23 Jun 2003 12:02:20 -0300 To: < > Subject: [c6x] Problem loading code with XDS560 Re: Hi, Re: Re: We are having problems to download the .out when using XDS560 emulator. Re: Re: We have a C6203 board and when we try to download the .out to the DSP via XDS560, a message appears warning that the endianess of the target is different from the endianess of the .out. The DSP was on-board programmed, via pull-up/pull-down, to work in little-endian, but when we read CSR register, via Code Composer, it's 8th bit is in '0', indicating big-endian. Re: Re: The DSP pin that programs the endianess is connected to a PLX PCI bridge pin, that drives the bus after the DSP reset. The DSP reads the pin value during its reset, so in normal functioning (without emulator), the endianess is correctly read by the DSP. But when we use the emulator, it seems that XDS560 reads the pin value to determine the endianess, but at this time the bus belongs to the PLX, that probably drives the pin to '0'. I think that the correct would be if XDS560 read the CSR value, not the pin value. Re: Re: Anyone could help me to solve this problem, i.e. what can I do in order to force the XDS560 to read correctly the endianess? Re: Re: Thank you, Re: Re: Regards, Re: Alexandre Re: Re: Re: Re: _____________________________________ Re: Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. Re: Re: _____________________________________ Re: About this discussion group: Re: Re: To Join: Send an email to Re: Re: To Post: Send an email to Re: Re: To Leave: Send an email to Re: Re: Archives: http://www.yahoogroups.com/group/c6x Re: Re: Other Groups: http://www.dsprelated.com Re: Re: Re: ">http://docs.yahoo.com/info/terms/ Re: Re: -- __________________________________________________________ Sign-up for your own FREE Personalized E-mail at Mail.com http://www.mail.com/?sr=signup CareerBuilder.com has over 400,000 jobs. Be smarter about your job search http://corp.mail.com/careers _____________________________________ Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. _____________________________________ About this discussion group: To Join: Send an email to To Post: Send an email to To Leave: Send an email to Archives: http://www.yahoogroups.com/group/c6x Other Groups: http://www.dsprelated.com ">http://docs.yahoo.com/info/terms/ |
Reply by ●June 24, 20032003-06-24
|
Alexandre Freire da Silva Osorio- > Thank you very much for your answers. > > The endianess pin in our board is already pulled-up (little-endian). > So, the problem seems to be the XDS560. It should read the 8th CSR > bit, not the pin directly. When we power-up the board, the DSP reads > the pin as logical "1", because PLX is not the bus master at this > time. After this, we cannot guarantee the pin value, because it is > being mastered by PLX. > > In respect to download a big-endian program, yes, we tried it and > had the same problem as Ganesh did. Many memory locations was not > correctly loaded. We tried it after we concluded that XDS would not > correctly read the DSP endianess. Yes Ganesh's and Tao's answers are very good, much better than mine. It sort of looks like CCS + XDS560 will not handle big endian yet, no matter what you try to do in hardware. Possible bug, maybe you could press TI for an answer. If you want, I can give you offline an engineering contact in TI's XDS560 test group. -Jeff > -----Original Message----- > From: Ganesh Vijayan [mailto:] > Sent: Tuesday, June 24, 2003 1:46 AM > To: Alexandre Freire da Silva Osorio; ; Tao Wang > Subject: Re: [c6x] Problem loading code with XDS560 > > Hi all, > Tao has given the right answer. When the pin is "pulled up", the board will be in little endian and when it is down it will be in big endian. > CSR can't be set by software as it is a read only bit. The 8th bit as Alexandre pointed out denotes the endianness. Hence when the power is switched on to the board, the status of the pin determines your endianness. > BTW I have a doubt. Are you trying a little endian program or a big endian program. I work on C6416 with same configuration but haven't had success with big endian board and XDS560. There seems to be some problem with XDS560 drivers. Actually for little endian mode, it works without a glitch but for big endian mode, it opens the CCS but when I try to load the program, it gives a problem with respect to memory mapping. Actually the GEL file provided with the XDS560 driver doesn't initialize the memories or rather memory-mapping doesn't occur. when I modify the same GEL file with all the memory mapping initializations, it doesn't open the CCS at all and hangs the CCS. I would like to know if anyone else has succeeded in loading and running a big endian program on the board with XDS560. If so, can you share the GEL file with me. > Also hope this clarifies your doubt. > Thanks and Regards, > Ganesh > > ----- Original Message ----- > From: Tao Wang <mailto:> > To: Alexandre Freire da Silva Osorio <mailto:> ; > Sent: Tuesday, June 24, 2003 6:49 AM > Subject: Re: [c6x] Problem loading code with XDS560 > > Hi, Alexandre Freire da Silva Osorio, > > The key problem is the time order of reset of dsp and the state of the endianess configuration pin. The pin must be high if you need little endian during the reset of dsp. Each time you reset(manual, power on) you dsp the pin is read and write to CSR. So why don't you directly pull it up? > > The endianess bit in CSR is read only. So it is not problem of software, it is your hardware. Have a scope to check the state of the configuration pin while the dsp is in reset. > > Wang Tao > > ----- Original Message ----- > From: "Alexandre Freire da Silva Osorio" < > > Date: Mon, 23 Jun 2003 12:02:20 -0300 > To: < > > Subject: [c6x] Problem loading code with XDS560 > > Re: Hi, > Re: > Re: We are having problems to download the .out when using XDS560 emulator. > Re: > Re: We have a C6203 board and when we try to download the .out to the DSP via XDS560, a message appears warning that the endianess of the target is different from the endianess of the .out. The DSP was on-board programmed, via pull-up/pull-down, to work in little-endian, but when we read CSR register, via Code Composer, it's 8th bit is in '0', indicating big-endian. > Re: > Re: The DSP pin that programs the endianess is connected to a PLX PCI bridge pin, that drives the bus after the DSP reset. The DSP reads the pin value during its reset, so in normal functioning (without emulator), the endianess is correctly read by the DSP. But when we use the emulator, it seems that XDS560 reads the pin value to determine the endianess, but at this time the bus belongs to the PLX, that probably drives the pin to '0'. I think that the correct would be if XDS560 read the CSR value, not the pin value. > Re: > Re: Anyone could help me to solve this problem, i.e. what can I do in order to force the XDS560 to read correctly the endianess? > Re: > Re: Thank you, > Re: > Re: Regards, > Re: Alexandre |
Reply by ●June 24, 20032003-06-24
|
Jeff, Yes, I would like very much to obtain this contact in XDS560 test group. Thank you. Alexandre -----Original Message----- From: Jeff Brower [mailto:] Sent: Tuesday, June 24, 2003 11:05 AM To: Alexandre Freire da Silva Osorio Cc: Subject: Re: [c6x] Problem loading code with XDS560 Alexandre Freire da Silva Osorio- > Thank you very much for your answers. > > The endianess pin in our board is already pulled-up (little-endian). > So, the problem seems to be the XDS560. It should read the 8th CSR > bit, not the pin directly. When we power-up the board, the DSP reads > the pin as logical "1", because PLX is not the bus master at this > time. After this, we cannot guarantee the pin value, because it is > being mastered by PLX. > > In respect to download a big-endian program, yes, we tried it and > had the same problem as Ganesh did. Many memory locations was not > correctly loaded. We tried it after we concluded that XDS would not > correctly read the DSP endianess. Yes Ganesh's and Tao's answers are very good, much better than mine. It sort of looks like CCS + XDS560 will not handle big endian yet, no matter what you try to do in hardware. Possible bug, maybe you could press TI for an answer. If you want, I can give you offline an engineering contact in TI's XDS560 test group. -Jeff > -----Original Message----- > From: Ganesh Vijayan [mailto:] > Sent: Tuesday, June 24, 2003 1:46 AM > To: Alexandre Freire da Silva Osorio; ; Tao Wang > Subject: Re: [c6x] Problem loading code with XDS560 > > Hi all, > Tao has given the right answer. When the pin is "pulled up", the board will be in little endian and when it is down it will be in big endian. > CSR can't be set by software as it is a read only bit. The 8th bit as Alexandre pointed out denotes the endianness. Hence when the power is switched on to the board, the status of the pin determines your endianness. > BTW I have a doubt. Are you trying a little endian program or a big endian program. I work on C6416 with same configuration but haven't had success with big endian board and XDS560. There seems to be some problem with XDS560 drivers. Actually for little endian mode, it works without a glitch but for big endian mode, it opens the CCS but when I try to load the program, it gives a problem with respect to memory mapping. Actually the GEL file provided with the XDS560 driver doesn't initialize the memories or rather memory-mapping doesn't occur. when I modify the same GEL file with all the memory mapping initializations, it doesn't open the CCS at all and hangs the CCS. I would like to know if anyone else has succeeded in loading and running a big endian program on the board with XDS560. If so, can you share the GEL file with me. > Also hope this clarifies your doubt. > Thanks and Regards, > Ganesh > > ----- Original Message ----- > From: Tao Wang <mailto:> > To: Alexandre Freire da Silva Osorio <mailto:> ; > Sent: Tuesday, June 24, 2003 6:49 AM > Subject: Re: [c6x] Problem loading code with XDS560 > > Hi, Alexandre Freire da Silva Osorio, > > The key problem is the time order of reset of dsp and the state of the endianess configuration pin. The pin must be high if you need little endian during the reset of dsp. Each time you reset(manual, power on) you dsp the pin is read and write to CSR. So why don't you directly pull it up? > > The endianess bit in CSR is read only. So it is not problem of software, it is your hardware. Have a scope to check the state of the configuration pin while the dsp is in reset. > > Wang Tao > > ----- Original Message ----- > From: "Alexandre Freire da Silva Osorio" < > > Date: Mon, 23 Jun 2003 12:02:20 -0300 > To: < > > Subject: [c6x] Problem loading code with XDS560 > > Re: Hi, > Re: > Re: We are having problems to download the .out when using XDS560 emulator. > Re: > Re: We have a C6203 board and when we try to download the .out to the DSP via XDS560, a message appears warning that the endianess of the target is different from the endianess of the .out. The DSP was on-board programmed, via pull-up/pull-down, to work in little-endian, but when we read CSR register, via Code Composer, it's 8th bit is in '0', indicating big-endian. > Re: > Re: The DSP pin that programs the endianess is connected to a PLX PCI bridge pin, that drives the bus after the DSP reset. The DSP reads the pin value during its reset, so in normal functioning (without emulator), the endianess is correctly read by the DSP. But when we use the emulator, it seems that XDS560 reads the pin value to determine the endianess, but at this time the bus belongs to the PLX, that probably drives the pin to '0'. I think that the correct would be if XDS560 read the CSR value, not the pin value. > Re: > Re: Anyone could help me to solve this problem, i.e. what can I do in order to force the XDS560 to read correctly the endianess? > Re: > Re: Thank you, > Re: > Re: Regards, > Re: Alexandre |
Reply by ●June 25, 20032003-06-25
|
Hi, Jeff and All, I am also use xds560 with dsk6711 under windows98. I have not meet the big endian problem because I use little endian. But I think there are other problem with the emulator. So please let me know if there is some test result. There is a .gel file for the dsk6711, it is very detailed, I don't know if it can be refereced. I have a problem with dsk6711. Hardware: 1.dsk6711 rev 2.0 with no daughterboard 2.xds560 emulator and JTAG head 3.Intel Celeron 400Mhz PC 4.Creative soundcard and active sound box Software: 1.Micrsoft Windows 98 2.CCS 2.1 I use JTAG to connect the dsk and PC. J7(in) is wired to the soundcard on PC. J6(out) is wired to the active sound box. I add "*(int *)0x01800004 = 0xFFFFFF03;" to the gel file 'dsk6211_6711.gel' to see contents of the flash with emulator. I revised the POST program with my own bootloader(post.prj). I also wrote a program that program the flash through JTAG and emulator(flash.prj). Under emulator the POST program(with lnk.cmd) work good. I relink the POST program with lnk_load.cmd and parse 'post.out' to 'code'(raw data in 32 bits). Then I program the flash with 'code' using the program I write(flash.prj). I close CCS and turn off the power and then turn on the power. But the system does not work. I reopen CCS and view the dissambly find that the PC stop at address 0x00000000. I run it step by step, it's OK. From the phenomenon we can see the dsp has bootloaded with copying the first 1k code from the flash to internal sram but does not run the program. The reset button and 'xdsreset' also has no use. Is the DSP6711 ruined or the DSK is ruined or there is something wrong with my program or I have done some misoperation? Or the power has some problem. The attachments: project 'post', project 'flash', and the revised "dsk6211_6711.gel". I have send my problem to ti asia, but they say they can not open .rar, they need .zip. I have to send it again with .zip. Wang Tao ----- Original Message ----- From: Jeff Brower <> Date: Tue, 24 Jun 2003 09:04:55 -0500 To: Alexandre Freire da Silva Osorio <> Subject: Re: [c6x] Problem loading code with XDS560 Re: Alexandre Freire da Silva Osorio- Re: Re: > Thank you very much for your answers. Re: > Re: > The endianess pin in our board is already pulled-up (little-endian). Re: > So, the problem seems to be the XDS560. It should read the 8th CSR Re: > bit, not the pin directly. When we power-up the board, the DSP reads Re: > the pin as logical "1", because PLX is not the bus master at this Re: > time. After this, we cannot guarantee the pin value, because it is Re: > being mastered by PLX. Re: > Re: > In respect to download a big-endian program, yes, we tried it and Re: > had the same problem as Ganesh did. Many memory locations was not Re: > correctly loaded. We tried it after we concluded that XDS would not Re: > correctly read the DSP endianess. Re: Re: Yes Ganesh's and Tao's answers are very good, much better than mine. It sort of Re: looks like CCS + XDS560 will not handle big endian yet, no matter what you try to do Re: in hardware. Possible bug, maybe you could press TI for an answer. If you want, I Re: can give you offline an engineering contact in TI's XDS560 test group. Re: Re: -Jeff Re: Re: > -----Original Message----- Re: > From: Ganesh Vijayan [mailto:] Re: > Sent: Tuesday, June 24, 2003 1:46 AM Re: > To: Alexandre Freire da Silva Osorio; ; Tao Wang Re: > Subject: Re: [c6x] Problem loading code with XDS560 Re: > Re: > Hi all, Re: > Tao has given the right answer. When the pin is "pulled up", the board will be in little endian and when it is down it will be in big endian. Re: > CSR can't be set by software as it is a read only bit. The 8th bit as Alexandre pointed out denotes the endianness. Hence when the power is switched on to the board, the status of the pin determines your endianness. Re: > BTW I have a doubt. Are you trying a little endian program or a big endian program. I work on C6416 with same configuration but haven't had success with big endian board and XDS560. There seems to be some problem with XDS560 drivers. Actually for little endian mode, it works without a glitch but for big endian mode, it opens the CCS but when I try to load the program, it gives a problem with respect to memory mapping. Actually the GEL file provided with the XDS560 driver doesn't initialize the memories or rather memory-mapping doesn't occur. when I modify the same GEL file with all the memory mapping initializations, it doesn't open the CCS at all and hangs the CCS. I would like to know if anyone else has succeeded in loading and running a big endian program on the board with XDS560. If so, can you share the GEL file with me. Re: > Also hope this clarifies your doubt. Re: > Thanks and Regards, Re: > Ganesh Re: > Re: > ----- Original Message ----- Re: > From: Tao Wang <mailto:> Re: > To: Alexandre Freire da Silva Osorio <mailto:> ; Re: > Sent: Tuesday, June 24, 2003 6:49 AM Re: > Subject: Re: [c6x] Problem loading code with XDS560 Re: > Re: > Hi, Alexandre Freire da Silva Osorio, Re: > Re: > The key problem is the time order of reset of dsp and the state of the endianess configuration pin. The pin must be high if you need little endian during the reset of dsp. Each time you reset(manual, power on) you dsp the pin is read and write to CSR. So why don't you directly pull it up? Re: > Re: > The endianess bit in CSR is read only. So it is not problem of software, it is your hardware. Have a scope to check the state of the configuration pin while the dsp is in reset. Re: > Re: > Wang Tao Re: > Re: > ----- Original Message ----- Re: > From: "Alexandre Freire da Silva Osorio" < > Re: > Date: Mon, 23 Jun 2003 12:02:20 -0300 Re: > To: < > Re: > Subject: [c6x] Problem loading code with XDS560 Re: > Re: > Re: Hi, Re: > Re: Re: > Re: We are having problems to download the .out when using XDS560 emulator. Re: > Re: Re: > Re: We have a C6203 board and when we try to download the .out to the DSP via XDS560, a message appears warning that the endianess of the target is different from the endianess of the .out. The DSP was on-board programmed, via pull-up/pull-down, to work in little-endian, but when we read CSR register, via Code Composer, it's 8th bit is in '0', indicating big-endian. Re: > Re: Re: > Re: The DSP pin that programs the endianess is connected to a PLX PCI bridge pin, that drives the bus after the DSP reset. The DSP reads the pin value during its reset, so in normal functioning (without emulator), the endianess is correctly read by the DSP. But when we use the emulator, it seems that XDS560 reads the pin value to determine the endianess, but at this time the bus belongs to the PLX, that probably drives the pin to '0'. I think that the correct would be if XDS560 read the CSR value, not the pin value. Re: > Re: Re: > Re: Anyone could help me to solve this problem, i.e. what can I do in order to force the XDS560 to read correctly the endianess? Re: > Re: Re: > Re: Thank you, Re: > Re: Re: > Re: Regards, Re: > Re: Alexandre Re: Re: _____________________________________ Re: Note: If you do a simple "reply" with your email client, only the author of this message will receive your answer. You need to do a "reply all" if you want your answer to be distributed to the entire group. Re: Re: _____________________________________ Re: About this discussion group: Re: Re: To Join: Send an email to Re: Re: To Post: Send an email to Re: Re: To Leave: Send an email to Re: Re: Archives: http://www.yahoogroups.com/group/c6x Re: Re: Other Groups: http://www.dsprelated.com Re: Re: Re: ">http://docs.yahoo.com/info/terms/ Re: Re: -- __________________________________________________________ Sign-up for your own FREE Personalized E-mail at Mail.com http://www.mail.com/?sr=signup CareerBuilder.com has over 400,000 jobs. Be smarter about your job search http://corp.mail.com/careers | |||
|
Reply by ●June 25, 20032003-06-25
|
Hi Jeff, You have hit the nail, right on the head. It definitely seems like CCS and XDS560 wont work together in Big Endian mode. We did try upgrading the CCS to 2.12, but haven't had much success. Also there is news that CCS 2.20 is coming with in-built XDS560 drivers. Does any one of you have CCS 2.20 ? If so, can you try using CCS 2.20 and XDS 560. Regarding the bug, yeah we can press TI. May be offline we can press for an answer. You, Tao, Alexandre and me can be in loop and press for the answer. With Warm Regards, Ganesh ----- Original Message ----- From: Jeff Brower Sent: Tuesday, June 24, 2003 7:34 PM Subject: Re: [c6x] Problem loading code with XDS560 Yes Ganesh's and Tao's answers are very good, much better than mine. It sort of looks like CCS + XDS560 will not handle big endian yet, no matter what you try to do in hardware. Possible bug, maybe you could press TI for an answer. If you want, I can give you offline an engineering contact in TI's XDS560 test group. -Jeff |
|
|
