FreeCalypso > hg > freecalypso-tools
comparison doc/Flash-boot-wa @ 462:efb93f3e4ac7
doc/Flash-boot-wa: explanatory article added
| author | Mychaela Falconia <falcon@freecalypso.org> |
|---|---|
| date | Sun, 10 Feb 2019 21:51:28 +0000 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| 461:10e168596dfd | 462:efb93f3e4ac7 |
|---|---|
| 1 There is a tiny (120 bytes SREC file) program called flash-boot-wa that was | |
| 2 written in the spring of 2017 for the purpose of working around a problem that | |
| 3 happened on only one first-batch FCDEV3B board and never happened on any other | |
| 4 board - but Murphy's law had it that this one troubled board just had to be the | |
| 5 one on which my (Mother Mychaela's) very initial development and bring-up work | |
| 6 was done. | |
| 7 | |
| 8 The defect exhibited on that one board was as follows: it had no problem booting | |
| 9 serially (fc-iram, fc-loadtool, fc-xram) and it had no problem booting from | |
| 10 flash in mode 0 (see the Flash-boot-modes article in the freecalypso-docs | |
| 11 repository), but booting from flash in mode 1 (the flash boot mode used by | |
| 12 FC Magnetite, which is our primary firmware) was troubled. The exact failure | |
| 13 mode and the root cause were never solved, but the issue most likely involved | |
| 14 the watchdog reset in some way (it occurs as part of flash boot mode 1 but not | |
| 15 in mode 0 and not in serial downloading), and because Calypso's FDP output goes | |
| 16 low during watchdog reset (or at least TI's CAL000 document says so), it is | |
| 17 plausible that the root cause involved the Spansion flash chip getting unhappy | |
| 18 as a result of being jerked with extra resets which don't meet its reset timing | |
| 19 requirements. | |
| 20 | |
| 21 Our new FCDEV3B V2 boards no longer use Calypso's FDP output (it is left | |
| 22 unconnected) and feature a new flash reset circuit of our own design that meets | |
| 23 the reset timing requirements of our Spansion flash chip, hence the flash boot | |
| 24 problem seen on that one FCDEV3B S/N 001 board is not expected to recur on any | |
| 25 of our current or future boards. However, our little flash-boot-wa program is | |
| 26 kept around: removing a previously-released 120-byte program for no good reason | |
| 27 is not the way of FOSS. | |
| 28 | |
| 29 This flash-boot-wa program is loaded serially via fc-iram; it disables the boot | |
| 30 ROM and jumps to address 0 (the opposite of what we do in compalstage for Mot | |
| 31 C1xx phones), thereby indirectly booting the made-for-boot-mode-1 firmware image | |
| 32 in the flash. The intended usage was as follows: | |
| 33 | |
| 34 fc-iram -h fcfam /dev/ttyXXX /opt/freecalypso/target-bin/flash-boot-wa.srec rvinterf | |
| 35 | |
| 36 It is also worth noting that fc-iram has been extended to support second program | |
| 37 invokation just like fc-xram (used in the invokation line above) just for this | |
| 38 peculiar use case. The flash-boot-wa.srec helper can also be booted via | |
| 39 fc-xram. |