GitLab

* Configure second SD controller to drive SDIO WiFi module and report via a
  HAL device
* Ensure in PostPowerOn that we don't try to set an unsupported voltage
* No need to dynamically select GetWriteProtect function
* Enable background operations

Version 0.62. Tagged as 'HAL_OMAP4-0_62'

* Add missing `.gitignore`

* Makefile now uses shared HAL makefile fragment
* Filenames converted to OS-agnostic form
* Remove PIC attributes from AREA declarations: since the code isn't
  position-independent, this is factually incorrect, and will produce
  false linker warnings when we link with libraries that are also non-PIC
* Replaced a couple of UAL instructions with their pre-UAL counterparts to
  reduce number of warnings

HAL_UARTFIFOSize was incorrectly returning the TX FIFO size, instead of
the number of characters that can be written when the TX empty IRQ is
received. This was causing data loss when DualSerial drives the
interface at high speed (> 19200 baud).

Fix this by programming a sensible trigger level at startup, and
returning the correct value from HAL_UARTFIFOSize.

Version 0.61. Tagged as 'HAL_OMAP4-0_61'

Detail:
  Makefile, s/DBell - Add doorbell HAL device implementation, using GIC SGIs
  hdr/StaticWS, hdr/omap4430 - Tidy things up a bit by removing the DebugInterrupts code.
  hdr/omap4_reg - Offset all the shared interrupt numbers so that RISC OS IRQs 0-31 are core 0 private, 32-63 core 1 private, 64-159 shared
  s/Boot - Implement SMP ARMops
  s/Interrupts - Tidy up interrupt handling and implement new HAL IRQ calls
Admin:
  Untested
  Requires Kernel-6_09


Version 0.60. Tagged as 'OMAP4-0_60'

Detail:
  Correct group number for pin 58.
  Use OMAP44x0 register names for interrupt handling, the status registers are split into set/clear pairs.
Admin:
  Submission from Willi Theiss.

Version 0.59. Tagged as 'OMAP4-0_59'

Update to API 1.0, and corresponding set of register wide accessor functions.
The tables of permitted pins come from staring at the schematics for any connectors, minus those that are "owned" by the OS.

Tested on a Pandaboard, toggling the two user LEDs and scanning the 'USER' push button (which, by the way, needs the pullup enabling as there isn't a discrete one).

Version 0.58. Tagged as 'OMAP4-0_58'

Detail:
  SDIO.s comment correction
  hdr/StaticWS removed unused USBAllocArea
Admin:
  Submission from Willi Theiss.

Version 0.57. Tagged as 'OMAP4-0_57'

Adopt the board type and revision enum privately in board.hdr (mass search & replace exercise).
Don't re-export it via a stub GPIO device; it's private.

Version 0.56. Tagged as 'OMAP4-0_56'

Detail:
  Added check for already running conversion in SR44x_GetDieTempES use continuous conversion mode in SR44x_GetDieTemp to reduce waiting phases.
Admin:
  Submission from Willi Theiss.

Version 0.55. Tagged as 'OMAP4-0_55'

Detail:
  Replace keyboard scan code with list of modules that the kernel needs to do the same.
  Reorder the HALEntries to match Kernel-5_89.
  Delete unused workspace.
  Don't link against HAL USB library.
Admin:
  Submission for USB bounty.

Version 0.54. Tagged as 'OMAP4-0_54'

Admin:
  As a side effect of changes in USBDriver-1_20, rename the library object.

Version 0.53. Tagged as 'OMAP4-0_53'

Detail:
  s/SDIO - Add a GET of Hdr:CPU.Arch, as it's now required for DivRem usage
Admin:
  Untested


Version 0.52. Tagged as 'OMAP4-0_52'

Detail:
  s/Boot, s/Interrupts, s/Timers - Move timer IRQ clear code out of HAL_IRQClear and into HAL_TimerIRQClear. Streamlines HAL_IRQClear, and necessary to avoid undefined symbols when linking with latest HAL USB library
Admin:
  Builds, untested


Version 0.51. Tagged as 'OMAP4-0_51'

Detail:
  Return the board name for this entry.
  Rename CPU temperature sensor register per datasheet.
Admin:
  Submission from Willi Theiss.
  Not tested.

Version 0.50. Tagged as 'OMAP4-0_50'

Detail:
  Extra entry to get the die temperature, as reported by Portable_ReadSensor(0,0,0).
Admin:
  Submission from Willi Theiss.
  Tested on OMAP4430 and OMAP4460.

Version 0.49. Tagged as 'OMAP4-0_49'

Detail:
  There's no need to provide an empty implementation, just use NullEntry.
Admin:
  Submission from Willi Theiss.

Version 0.48. Tagged as 'OMAP4-0_48'

omap4430.hdr/Debug,s:
  Define a dummy symbol so when debug is enabled there aren't conflicting IMPORT and EXPORTs.
PRCM.s:
  Fill in missing frequency.
Boot.s:
  Update comment to match HAL function names.
NVMemory.s:
  Add offset to account for FATLoadedCMOS not being at offset 0.
Top.s:
  Remove unused Post header. Replace DCI with instruction now objasm supports it.
Submission from Willi Theiss.

Version 0.47. Tagged as 'OMAP4-0_47'

Detail:
  s/SDMA - Add extra memory barriers to DMA code, ready for when the default NCB memory cache policy is changed from Device to Normal, non-cacheable (increased risk of out-of-order accesses)
  s/PL310 - Add DMB ST implementation that also deals with PL310 synchronisation
Admin:
  Tested on PandaBoard


Version 0.46. Tagged as 'OMAP4-0_46'

Detail:
  s/RTC - Remove TWL6030 RTC startup code from RTC_Init, to ensure that RTC_Read is able to report that the time is invalid if the RTC isn't running. This avoids the RTC module getting confused and thinking the century needs incrementing (RTC will report a year of xx00, CMOS will hold 2015, RTC module assumes year can only go forward -> therefore must be 2100)
  Also perform a dummy read of RTC_STATUS_REG before reading actual value, as the manual warns that this is required.
Admin:
  Tested on Pandaboard
  With CMOS available:
  - *Time defaults to year 1970 on power up (CMOS year ignored due to RTC read error)
  - After setting time and rebooting, time is retained and 4 digit year is reconstructed correctly
  Without CMOS available:
  - *Time defaults to year 1970 on power up (default CMOS year ignored due to RTC read error)
  - After setting time and rebooting, time is retained and 4 digit year is reconstructed correctly (RTC year xx15 deemed higher than CMOS default year 2000)


Version 0.45. Tagged as 'OMAP4-0_45'

Anyone using the serial port for other peripheral uses might get confused if some HAL debug comes out on it.
Move the IMPORTs into a header so we don't end up with lots of unresolved symbols with Debug {FALSE}.
Built, but not tested.

Version 0.44. Tagged as 'OMAP4-0_44'

Updated for Sound0HAL-1_80.

Version 0.43. Tagged as 'OMAP4-0_43'

Detail:
  s/PL310 - Implemented PL310 disable & reset entries, to allow the L2 cache to be turned off as well as on. Fix potential register corruption from Activate entry (OMAP ROM code may corrupt r0-r12, lr)
Admin:
  Tested on Pandaboard
  Fixes crash on *Cache Off (in conjunction with Kernel-5_35-4_79_2_254)


Version 0.42. Tagged as 'OMAP4-0_42'

Detail:
  s/PL310 - Add basic HAL device for the PL310 cache controller. Reports the base address to the OS and handles basic cache initialisation.
  Makefile - Include PL310 code
  hdr/StaticWS - Allocate workspace for HAL device
  s/Boot - Remove L2 cache initialisation from HAL_Init. Register PL310 device during HAL_InitDevices.
Admin:
  Tested on rev A6 Pandaboard
  Requires Kernel-5_35-4_79_2_252


Version 0.41. Tagged as 'OMAP4-0_41'

On the OMAP4 we're in a privileged but unsecure mode, so to do the initial PL310 enable we call an SMC sunction embedded in the OMAP's onboard mask ROM to do the deed.
Also, correct the range of SDRAM hinted to the kernel - this looks like a copy and paste mistake from the Tungsten HAL.

Version 0.40. Tagged as 'OMAP4-0_40'

According to Willi, although MMC1 is connected at L3, RISC OS only ever uses the L4_PER interface, so move it back to that bus.
Retagged as OMAP4-0_39.

Note, the SDIO controller was previously misreported as being on L4, when in fact it's on L3.

Version 0.39. Tagged as 'OMAP4-0_39'

HAL_HardwareInfo - as word 0 specified that no IOMD or VIDC was present, don't waste time reporting anything in word 2.
Not tagged.

Version 0.38. Tagged as 'OMAP4-0_38'

As the flag bit in HAL_USBControllerInfo isn't set, no need to provide them.
Use symbols from Hdr:HALEntries when building the USB info.
No need for AASMFLAGS in the makefile.
Built, but not tested.

Version 0.37. Tagged as 'OMAP4-0_37'

Detail:
  hdr/PRCM:
  - added further definitions (not listed in older TRMs)
  s/Audio:
  - configure all GPIO pads used by audio system
    (for supporting latest U-Boot 2014-7)
  s/Boot:
  - added overflow check for address range
  s/RAM:
  - added overflow check for address range
  - address comparison with unsigned arithmetic
  s/RTC:
  - added initialisation code and pad configuration for TWL6030 RTC
    (for supporting latest U-Boot 2014-07)
  s/SDIO:
  - added pad configuration for LEDs used by SDIO
    (for supporting latest U-Boot 2014-07)
  s/USB:
  - removed unused code parts
  - use defines for controller types
  s/Video:
  - added pad configuration for DVI framer pin
  - enable DSS power and clock domain and DSS specific clocks
  - activate thermal sensor clock
  - wait for activated module in VideoDevice_Activate
    (for supporting latest U-Boot 2014-07)
Admin:
  Submission from Willi Theiss
  Tested on Panda ES rev B3 (Willi), Panda rev A6 (myself)


Version 0.36. Tagged as 'OMAP4-0_36'

Detail:
  s/I2C - Revise HAL_VideoIICOp implementation to disallow EDID writes (&A0 IIC address). Attempt to write the EDID segment pointer when accessing EDID addresses >= 256.
Admin:
  Builds but untested
  Same code as OMAP3-0_97


Version 0.35. Tagged as 'OMAP4-0_35'

hdr/Audio: split AudioChannels to allow new AudioDMASize field to differentiate OMAP4430 v 4460.
Audio.s: handle FIFO underflow events of the audio stream, differentiate FIFO sizes for OMAP4430 v 4460.
SDMA.s: add prefetch flag in DMA control for audio channel
Submission from Willi Theiss.

Version 0.34. Tagged as 'OMAP4-0_34'

Version 0.33. Tagged as 'OMAP4-0_33'

For a spurious interrupt the return value was not set to -1.
Also, the return value in the non spurious case was only correct because the CPU id is 0.
Inspected, but not tested.

Version 0.32. Tagged as 'OMAP4-0_32'

Avoids stalling of the interrupt system in some circumstances, in particular when shutting down the computer.
Submission from Willi Theiss.
Built, but not tested here.

Version 0.31. Tagged as 'OMAP4-0_31'

FIQ's aren't currently supported, but this is now controlled by the 'FIQ_Supported' switch.

Version 0.30. Tagged as 'OMAP4-0_30'

hdr.SDRC: Extra bit fields added to definitions
s.RAM: scan through all LISA_MAP register
       rework of clear loop (only one instance used)
s.Boot: scan through all LISA_MAP register

This should remove the dependence of the OMAP4 HAL to be run with a specific version of the accompanying loader.
Submission from Willi Theiss. Built, but not tested here,

Version 0.29. Tagged as 'OMAP4-0_29'

Objasm 4 supports ARMv7 opcodes, and this HAL is only every for ARMv7.
Built, but not tested. Submission from Willi Theiss.

Version 0.28. Tagged as 'OMAP4-0_28'

Detail:
  hdr/Video - Remove unused local VIDC list type 3 definition
  s/Boot, s/Video - Removed obsolete HAL video API implementation. A stub implementation is no longer required for systems that have a GraphicsV driver in a module, and can even cause problems if the OS decides to use the HAL implementation instead of the module one.
Admin:
  Builds, but untested. Essentially the same changes as for the OMAP3 HAL.
  Requires Kernel-5_35-4_79_2_203


Version 0.27. Tagged as 'OMAP4-0_27'