Merge request https://gitlab.rtems.org/rtems/rtos/rtems/-/merge_requests/1434 
was reviewed by Pavel Pisa

--
  
Pavel Pisa started a new discussion: 
https://gitlab.rtems.org/rtems/rtos/rtems/-/merge_requests/1434#note_157954


I have run updated RTEMS and LwIP `netstart01`/DHCP test successfully on 
`tms570ls3137_hdk_sdram`. I have modified it to stay running and run two ping 
floods against TMS570LS3137 LwIP from my laptop.

I have build `tms570ls3137_hdk` and `tms570lc4357_hdk` versions without 
Flashing yet and I have run `tms570lc4357_hdk_intram` with enabled 
initialization which has reached obtain of DHCP only sometimes. I expect that 
this can be for some uninitialized internal RAM memory (which is not fully 
synced because code is there) or some watchdog. I need to move forward with 
TM570LC4357 OpenOCD. But code I have been able to ping TMS570LC4357 from my 
laptop when `netstart01` test termination has been replaced by 
`rtems_task_suspend(RTEMS_SELF)`.

But generally I think that changes go to the the right direction.

--
  
Pavel Pisa commented on a discussion on 
spec/build/bsps/arm/tms570/optboardlchxl2.yml: 
https://gitlab.rtems.org/rtems/rtos/rtems/-/merge_requests/1434#note_157955

 > +enabled-by: true
 > +links: []
 > +name: TMS570_BOARD_LCHXL2_570LC43

We need two mapping for official boards and I would suggest to keep board names 
as individual defines, because they can be added in future. When none is 
selected then HDK is the default.

The idea about pin list and their use in the RTEMS system build only in the 
board respective HW initialization file (i.e.: `hwinit-lc4357-hdk.c` for 
TMS570LC4357 and `hwinit-ls3137-hdk.c` for TMS5703137) is such that when 
application runs on the custom board then `hwinit-XXXX.o` object file from 
`librtemsbsp.a` would be overridden by the application/user custom file 
providing the same set of basic functions and this way it would be possible to 
choose different PLLs setup, different peripherals mapping etc. without need to 
have separate RTEMS system build for each particular board.

In such case user can copy `hwinit-XXXX.c` prepare custom pins initialization 
list with even different name and use it for specific mapping. Even for HDK, 
when it is connected to some set of interfaces and peripherals, specific 
mapping is needed. So the default mapping was meant same basic setup to get run 
and then alter it in application specific manner.

The `xxx_PINMMR_DEFAULT_INIT_LIST` or some similar fragments can be recorded in 
application as `tms570_pinmmr_init_data` to make list minimal and allows 
collision free switch at large by direct MMR registers update.

What is missing in TMS570LC4357 is correct registers based update of MMRs 80 
till 99 for alternate inputs.

I am not sure if MII selects should or should not be excluded from the list.

For TMS570LC3137, the `TMS570LS3137ZWT_PINMMR_EMAC_MII_LIST` is not included in 
`TMS570_PINMMR_DEFAULT_INIT_LIST`.

For TMS570LC4257 the `TMS570LC4357_PINMMR_EMAC_MII_HDK_LIST` is included in 
`TMS570LC4357_PINMMR_DEFAULT_INIT_LIST` so this is probably mistake or at least 
the MII list should be added to TMS570LC3137 defaults too.

--
  
Pavel Pisa started a new discussion on 
bsps/arm/tms570/include/bsp/tms570lc4357-pins.h: 
https://gitlab.rtems.org/rtems/rtos/rtems/-/merge_requests/1434#note_157956

 > + * default terminals at alternative 0.  They carry no other function, so
 > + * there is no output PINMMR entry to set.
 > + *

Good catch, I expect that we have skipped the pins by scripts building initial 
lists because they have not been in main output mux and MMR table.

--
  
Pavel Pisa started a new discussion on 
bsps/arm/tms570/include/bsp/tms570lc4357-pins.h: 
https://gitlab.rtems.org/rtems/rtos/rtems/-/merge_requests/1434#note_157957

 > + * there is no output PINMMR entry to set.
 > + *
 > + * MDCLK and MII_TXD/MII_TXEN have no entry here.  They are outputs and

Good catch, we have the most probably did not notice that pins are not 
referenced from the main multiplexing/output table when we prepared the scripts 
to generate initial lists.

--
  
Pavel Pisa started a new discussion on 
bsps/arm/tms570/include/bsp/tms570lc4357-pins.h: 
https://gitlab.rtems.org/rtems/rtos/rtems/-/merge_requests/1434#note_157958

 > + * default terminals at alternative 0.  They carry no other function, so
 > + * there is no output PINMMR entry to set.
 > + *

Good catch, we have the most probably did not notice that pins are not 
referenced from the main multiplexing/output table when we prepared the scripts 
to generate initial lists.


-- 
View it on GitLab: 
https://gitlab.rtems.org/rtems/rtos/rtems/-/merge_requests/1434
You're receiving this email because of your account on gitlab.rtems.org. 
Unsubscribe from this thread: 
https://gitlab.rtems.org/-/sent_notifications/5-92akce5au77qm98kkn06q7yip-1d/unsubscribe
 | Manage all notifications: https://gitlab.rtems.org/-/profile/notifications | 
Help: https://gitlab.rtems.org/help


_______________________________________________
bugs mailing list
[email protected]
http://lists.rtems.org/mailman/listinfo/bugs

Reply via email to