Re: [U-Boot-Users] Can I write to flash using mw command
I did
protect off all
I have tried the following command sequences but didnt succeed in my objective of writing to and erasing first sector on flash.
WRITE COMMAND SEQUENCE:
mw ff000555 00AA00AA mw ff0002AA 00550055 mw ff000555 00A000A0 mw ff000000 12345678
SECTOR ERASE COMMAND SEQUENCE:
mw ff000555 00AA00AA mw ff0002AA 00550055 mw ff000555 00800080 mw ff000555 00AA00AA mw ff0002AA 00550055 mw ff000000 00300030
Note: I tried mw.b, mw.w, mw.l with appropriate data but didnt succeed.
In first place is it possible to write to AMD flash manually like this or shud it be done only programatically. Any one has tried writing to flash this way before ( I could write to Intel strata flash this way).
I walked through u-boot code for cfi_flash.c and flash_write() function that is called when cp (copy) command is invoked, the code does the same thing, the only extra thing it does is it polls on data register bits for a certain amount of time (flash erase, write timeout intervals) and also does a reset to bring back device to read array mode
mw.w ff000000 00F0 ; reset cmd
I have a ppc 8343 processor on board. I am wondering if U-boot that came with my target board has configuration that is preventing me to write to flash (but cp commands works fine with flash). Any one who has same spansion flash chip s29GL064M can try it, I cannot figure out the missing link in the chain, any advise, comments, suggestion will be of great help
Thanks, Bilahari
--- Wolfgang Denk wd@denx.de wrote:
In message
20061005222823.71219.qmail@web33206.mail.mud.yahoo.com
you wrote:
-> I read my board's spec and found out that I
have
two spansion s29GL064M devices on my board whose
data
bus is 16-bit wide and address bus is 32-bit wide. They are attached as 16-bit devices.
Then your command sequence was wrong.
-> I looked at the data sheet of the spansion s29GL64M and I am sure I am using the right command
sequences
(x16 Mode commands) for erase and program.
No. In 16 bit mode addresses get shifted by one bit, and your command word must address both chips, i. e. instead of
mw.b 0xff000555 0xAA
you need
mw FF000AAA 00AA00AA
-> How can I check if my memory controller is properly configured or not.
By carefully studying the hardware documentation, processor User's manual, chip documentation, etc. Then perform a code review.
Best regards,
Wolfgang Denk
-- Software Engineering: Embedded and Realtime Systems, Embedded Linux Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@denx.de Do you suppose the reason the ends of the `Intel Inside' logo don't match up is that it was drawn on a Pentium?
__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com
In message 20061006033619.65376.qmail@web33205.mail.mud.yahoo.com you wrote:
I did
protect off all
For this to work, you need working flash driver support in U-Boot.
And *if* you have working flash driver support, I wonder why you need to do such things:
WRITE COMMAND SEQUENCE:
mw ff000555 00AA00AA mw ff0002AA 00550055 mw ff000555 00A000A0 mw ff000000 12345678
Please explain which flash driver you are using (cfi driver???), and what the exact problem is you are trying to solve.
timeout intervals) and also does a reset to bring back device to read array mode
mw.w ff000000 00F0 ; reset cmd
In a configuration with 2 x 16 bit devices that should be 00F000F0.
I have a ppc 8343 processor on board. I am wondering if U-boot that came with my target board has configuration that is preventing me to write to flash (but cp commands works fine with flash). Any one who
This statement makes no sense to me. If cp to flash works fine, then obviously writing to flash is working fine, because this is what cp does when the target address points to flash memory - programming data to flash.
Explain what you want to do, and why, and what is not working - it sounds as if you have a perfectly working CFI driver on your board, so what is the problem?
Best regards,
Wolfgang Denk
bilahari akkiraju wrote:
I did
protect off all
I have tried the following command sequences but didnt succeed in my objective of writing to and erasing first sector on flash.
WRITE COMMAND SEQUENCE:
mw ff000555 00AA00AA mw ff0002AA 00550055 mw ff000555 00A000A0 mw ff000000 12345678
SECTOR ERASE COMMAND SEQUENCE:
mw ff000555 00AA00AA mw ff0002AA 00550055 mw ff000555 00800080 mw ff000555 00AA00AA mw ff0002AA 00550055 mw ff000000 00300030
Note: I tried mw.b, mw.w, mw.l with appropriate data but didnt succeed.
In first place is it possible to write to AMD flash manually like this or shud it be done only programatically. Any one has tried writing to flash this way before ( I could write to Intel strata flash this way).
I walked through u-boot code for cfi_flash.c and flash_write() function that is called when cp (copy) command is invoked, the code does the same thing, the only extra thing it does is it polls on data register bits for a certain amount of time (flash erase, write timeout intervals) and also does a reset to bring back device to read array mode
mw.w ff000000 00F0 ; reset cmd
I have a ppc 8343 processor on board. I am wondering if U-boot that came with my target board has configuration that is preventing me to write to flash (but cp commands works fine with flash). Any one who has same spansion flash chip s29GL064M can try it, I cannot figure out the missing link in the chain, any advise, comments, suggestion will be of great help
Thanks, Bilahari
You can write to flash by hand, if you do it right. It is a good way to learn how to do it programatically.
I believe the addresses you are using are wrong because your example shows two 16 bit chips to make up your 32 bits of data. This means that the LSB address line becomes a chip select between the two chips (logically, if not physically) and is not seen by the chips themselves. The result is the chips see the addresses shifted right 1 bit. Equivalently, you have shift your addresses left by 1 bit to compensate.
Chip address: xxx555 => 0101 0101 0101 Your address: xxxAAA => 101 0101 0101x (x is the dropped bit) Or, regrouped: 1010 1010 101o (filling in 'o' for the "unused" LSB)
Chip address: xxx2AA => 0010 1010 1010 Your address: xxx554 => 0001 0101 010x Or, regrouped: 0001 0101 010o (filling in 'o' for the "unused" LSB)
Try the following, I suspect it will work. WRITE COMMAND SEQUENCE:
mw ff000AAA 00AA00AA mw ff000554 00550055 mw ff000AAA 00A000A0 mw ff000000 12345678
The standard flash driver figures this all out for you and thus works.
gvb
Hi, Jerry,
Took your advice on flash addresses I should put into my command sequences with my current flash configuration (2 16-bit spansion s29GL064M chips interleaved to form 32-bit flash system). You pointed me to the right direction, thanks. Although I couldnt succeed in manual manipulation of flash, I am able to do it programatically (I downloaded a free low level driver provided by spansion and used it with my code). And it works fine, just like u-boot driver (cfi_flash.c). For anyone interested here is the link http://www.spansion.com/support/drivers_software/index.html
But one thing that doesnt work with my code and also with u-boot is that if I want to write two different values to the same flash locationconsecutively ( for example when I try to write 0x12 to flash address 0xff000000 and then write 0x34 to same flash address, u-boot gives me a message "flash not erased". My driver makes the program simply loop forever in a status register polling loop, I guess).
Is it something natural with flash programming ?. Should I always erase the whole sector before I re-program a location ? (doesnt sound like a good approach to me), is it not possible to write consecutively as with the conventional memory?
Thanks for the help and the directions
Bilahari
Bilahari,
On Wed, 2006-10-11 at 22:17 +0000, bilahari wrote:
But one thing that doesnt work with my code and also with u-boot is that if I want to write two different values to the same flash locationconsecutively ( for example when I try to write 0x12 to flash address 0xff000000 and then write 0x34 to same flash address, u-boot gives me a message "flash not erased". My driver makes the program simply loop forever in a status register polling loop, I guess).
Is it something natural with flash programming ?. Should I always erase the whole sector before I re-program a location ? (doesnt sound like a good approach to me), is it not possible to write consecutively as with the conventional memory?
Flash can't be accessed like the RAM you're thinking of. You can only change a bit state from 1 to 0 (That's why after erasing a sector you'll see that the contents are all 'ffffffff'). For all intents and purposes, once you write to a location in flash, it's 'read-only' until you erase the whole sector again. Also, keep in mind that flash devices have a limited number of erase cycles before performance becomes unreliable. A common value is 100000 cycles, although there are devices with more and less than this number. In general, use Flash to store information that will change infrequently.
regards, Ben
bilahari wrote:
Hi, Jerry,
Took your advice on flash addresses I should putinto my command sequences with my current flash configuration (2 16-bit spansion s29GL064M chips interleaved to form 32-bit flash system). You pointed me to the right direction, thanks. Although I couldnt
I guess Jerry guessed right how your flash chips are interfaced to the CPU bus since you did not disclose the complete information before. Without this information you cannot program the flash correctly.
But one thing that doesnt work with my code and also with u-boot is that if I want to write two different values to the same flash locationconsecutively ( for example when I try to write 0x12 to flash address 0xff000000 and then write 0x34 to same flash address, u-boot gives me a message "flash not erased". My driver makes the program simply loop forever in a status register polling loop, I guess).
Is it something natural with flash programming ?.
Yes. With NOR flash devices erasing turns flash to all 1s. By write you can turn 1s to 0s. But you cannot turn 0s to 1s without doing an erase of the erase unit (block/sector).
Should I always erase the whole sector before I re-program a location ? (doesnt sound like a good approach to me),
It does not sound that you have experience in embedded devices in particular with flash parts. I suggest you do some reading to familiarize yourself with the devices on your board. BTW, all this is really not U-Boot related and off topic on the list.
is it not possible to write consecutively as with the conventional memory?
It works as long as do not try to turn 1 bits to 0.
Best regards, Tolunay
Tolunay Orkun wrote:
is it not possible to write consecutively as with the conventional memory?
It works as long as do not try to turn 1 bits to 0.
Correction: as long as you do not try to turn 0 bits to 1.
Tolunay
Tolunay Orkun wrote:
Tolunay Orkun wrote:
is it not possible to write consecutively as with the conventional memory?
It works as long as do not try to turn 1 bits to 0.
Correction: as long as you do not try to turn 0 bits to 1.
Tolunay
Back when I was a kid, all we had were ones. To make a zero we had to subtract two ones. To change a zero to a one, we had to divide by zero to get an infinite number and then divide it by itself again to get a one... its no wonder flash wears out!
gvb ;-)
In message 452E2DD0.3070409@smiths-aerospace.com you wrote:
Back when I was a kid, all we had were ones. To make a zero we had to
They changed it to zeros later, because a zero is much more powerful. For example, leading zeros can be very powerful, expecially when you employ them in the management of big companies. [BTW: that's the difference between commercial and scientific notation: scientists tend to suppress leading zeros.] And, to quote an ancient usenet posting (sorry, the exact source is unknown to me):
Zero is an enigmatic value. It can mean success (fclose) or failure (scanf). It can mean black or white. It can mean no permissions (chmod) or all permissions (umask). It can mean now (setjmp) or later (atexit). It can mean the beginning (lseek) or the end (read). It can mean myself (getpgrp) or child (fork). It can mean all (kill's 1st argument) or nothing (kill's 2nd argument). It can mean `default' (SIG_IGN) or `I don't care' (waitpid) or `try to guess' (strtol). Indeed 0 lets you talk to God (setuid). Verily is 0 all things to all people.
SCNR :-)
Best regards,
Wolfgang Denk
On Thu, 2006-10-12 at 15:11 +0200, Wolfgang Denk wrote:
In message 452E2DD0.3070409@smiths-aerospace.com you wrote:
Back when I was a kid, all we had were ones. To make a zero we had to
They changed it to zeros later, because a zero is much more powerful. For example, leading zeros can be very powerful, expecially when you employ them in the management of big companies. [BTW: that's the difference between commercial and scientific notation: scientists tend to suppress leading zeros.] And, to quote an ancient usenet posting (sorry, the exact source is unknown to me):
Zero is an enigmatic value. It can mean success (fclose) or failure (scanf). It can mean black or white. It can mean no permissions (chmod) or all permissions (umask). It can mean now (setjmp) or later (atexit). It can mean the beginning (lseek) or the end (read). It can mean myself (getpgrp) or child (fork). It can mean all (kill's 1st argument) or nothing (kill's 2nd argument). It can mean `default' (SIG_IGN) or `I don't care' (waitpid) or `try to guess' (strtol). Indeed 0 lets you talk to God (setuid). Verily is 0 all things to all people.
Obligatory follow-on from a long-forgotten source:
"The decline of the Roman Empire can be traced to the fact that, lacking zero, they were unable to return from their C programs"
Ben Warren <bwarren <at> qstreams.com> writes:
On Thu, 2006-10-12 at 15:11 +0200, Wolfgang Denk wrote:
In message <452E2DD0.3070409 <at> smiths-aerospace.com> you wrote:
Back when I was a kid, all we had were ones. To make a zero we had to
They changed it to zeros later, because a zero is much more powerful. For example, leading zeros can be very powerful, expecially when you employ them in the management of big companies. [BTW: that's the difference between commercial and scientific notation: scientists tend to suppress leading zeros.] And, to quote an ancient usenet posting (sorry, the exact source is unknown to me):
Zero is an enigmatic value. It can mean success (fclose) or failure (scanf). It can mean black or white. It can mean no permissions (chmod) or all permissions (umask). It can mean now (setjmp) or later (atexit). It can mean the beginning (lseek) or the end (read). It can mean myself (getpgrp) or child (fork). It can mean all (kill's 1st argument) or nothing (kill's 2nd argument). It can mean `default' (SIG_IGN) or `I don't care' (waitpid) or `try to guess' (strtol). Indeed 0 lets you talk to God (setuid). Verily is 0 all things to all people.
Obligatory follow-on from a long-forgotten source:
"The decline of the Roman Empire can be traced to the fact that, lacking zero, they were unable to return from their C programs"
Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&da...
Thank you all , learned the The Transcedental Truth about zero,
From spansion documentation :-
<<
When Flash memory is erased, all the bits in the erased area are set to a logical “1” (high signal level). Programming selectively changes bits to a logical “0” (Low signal level). Flash memory is traditionally programmed a single byte or word at a time and erased one sector at a time. Only an erase operation can return a “0” back to a “1”(erased) state.
For Floating-Gate technology, it is possible to re-program (or write over) a byte or word that has been previously programmed without first erasing the byte or word (as long as no attempt is made to program a “0” to a “1”). This re-programming of byte or word locations can be used to incrementally program individual bits to “0.” This incremental programming is useful to track the progress in multi-step programming schemes such as those used in a Flash File System.
participants (6)
-
Ben Warren -
bilahari -
bilahari akkiraju
-
Jerry Van Baren -
Tolunay Orkun -
Wolfgang Denk