GitLab

Improve support for VMSAv6 cache policies & memory types. Expose raw ARMops via OS_MMUControl & cache information via OS_PlatformFeatures.

Detail:
  Docs/HAL/ARMop_API - Document two new ARMops: Cache_Examine and IMB_List
  hdr/KernelWS - Shuffle workspace round a bit to allow space for the two new ARMops. IOSystemType now deleted (has been deprecated and fixed at 0 for some time)
  s/ARM600 - Cosmetic changes to BangCam to make it clearer what's going on. Add OS_MMUControl 2 (get ARMop) implementation.
  s/ARMops - Switch out different ARMop implementations and XCB tables depending on MMU model - helps reduce assembler warnings and make it clearer what code paths are and aren't possible. Add implementations of the two new ARMops. Simplify ARM_Analyse_Fancy by removing some tests which we know will have certain results. Use CCSIDR constants in ARMv7 ARMops instead of magic numbers. Update XCB table comments, and add a new table for VMSAv6
  s/ChangeDyn - Define constant for the new NCB 'idempotent' cache policy (VMSAv6 normal, non-cacheable memory)
  s/HAL - Use CCSIDR constants instead of magic numbers. Extend RISCOS_MapInIO to allow the TEX bits to be specified.
  s/Kernel - OS_PlatformFeatures 33 (read cache information) implementation (actually, just calls through to an ARMop)
  s/MemInfo - Modify VMSAv6 OS_Memory 0 cache/uncache implementation to use the XCB table instead of modifying L2_C directly. This allows the cacheability to be changed without affecting the memory type - important for e.g. unaligned accesses to work correctly. Implement cache policy support for OS_Memory 13.
  s/Middle - Remove IOSystemType from OS_ReadSysInfo 6.
  s/VMSAv6 - Make sure BangCam uses the XCB table for working out the attributes of temp-uncacheable pages instead of manipulating L2_C directly. Add OS_MMUControl 2 implementation.
  s/AMBControl/memmap - Update VMSAv6 page table pokeing to use XCB table
  s/PMF/osinit - Remove IOSystemType reference, and switch out some pre-HAL code that was trying to use IOSystemType.
Admin:
  Tested on Iyonix, ARM11, Cortex-A7, -A8, -A9, -A15
  Note that contrary to the comments in the source the default NCB policy currently maps to VMSAv6 Device memory type (as per previous kernel versions). This is just a temporary measure, and it will be switched over to Normal, non-cacheable once appropriate memory barriers have been added to the affected IO code.


Version 5.35, 4.79.2.273. Tagged as 'Kernel-5_35-4_79_2_273'

Expose more areas via OS_ReadSysInfo 6 & OS_Memory 16. Expose processor vectors base + size via OS_PlatformFeatures.

Detail:
  hdr/KernelWS - Define processor vectors address. Currently same as ZeroPage, but in the future will differ for some machines.
  hdr/OSRSI6, s/Middle - Expose VecPtrTab & NVECTORS via OS_ReadSysInfo items 85 & 86
  s/Kernel - Add OS_PlatformFeatures 32, for returning the base + size of the processor vectors
  s/MemInfo - Add areas 12 thru 15 to OS_Memory 16, for reporting ZeroPage, ProcVecs, DebuggerSpace and ScratchSpace. The task manager can now use these for calculating memory usage instead of assuming 32K workspace from &0-&8000.
Admin:
  Tested on Raspberry Pi


Version 5.35, 4.79.2.271. Tagged as 'Kernel-5_35-4_79_2_271'

Detail:
  s/ARMops - Implement Cache_RangeThreshold for PL310 (helps AMBControl to decide what type of TLB maintenance is best). Fix MMU_ChangingEntry_PL310 doing more work than is necessary; was attempting to flush all ways for a given address tag, when really it should have only been flushing all the lines within a page and letting the cache worry about the tags/indices they correspond to.
  s/ChangeDyn, s/VMSAv6, s/AMBControl/memmap - Do extra TLB maintenance following writes to the page tables, as mandated by the ARMv6+ memory order model. Fixes frequent crashes on Cortex-A9 when running with lazy task swapping disabled (and presumably fixes other crashes too)
  s/MemInfo - Fix OS_Memory cache/uncache so that it does cache/TLB maintenance on a per-page basis instead of a global basis. Vastly improves performance when you have a large cache, but may need tweaking again in future to do a global op if large numbers of pages are being modified.
Admin:
  Tested on Pandaboard


Version 5.35, 4.79.2.255. Tagged as 'Kernel-5_35-4_79_2_255'

MemInfo.s:
 Several places in this code called the HAL or other ATPCS defined functions like memset() and hoped that the overall result was V clear. If any of them accidentally set V (for example a CMP that straddles 0x80000000) you ended up trying to look up an international error at the address of the reason code to OS_Memory.
 Now, explicitly clear V in the non error cases where an ATPCS function was called.
 Change the HAL_PhysInfo call to expect a physical ROM size back as an inclusive range, to match the RAM range subreason code. Add 1 to correct for this. A value of 0 & 0 is taken to mean "no physical ROM" as before.
Middle.s:
 Document that 255 means "no IOMD" or "no VIDC", that's what the HALs have been using since year dot.

Version 5.35, 4.79.2.248. Tagged as 'Kernel-5_35-4_79_2_248'

Detail:
  s/MemInfo - Fixed typo causing build error with HiProcVecs/zero page relocated kernel
  s/PMF/osinit - Fix the call to GraphicsV_StartupMode to work correctly with non-zero driver numbers
Admin:
  Tested on BB-xM with high processor vectors


Version 5.35, 4.79.2.226. Tagged as 'Kernel-5_35-4_79_2_226'

Add OS_Memory 24 implementation. Change OS_ValidateAddress to use it. Fix kernel leaving the physical access MB in a messy state. Try and protect against infinite abort loops caused by bad environment handlers.

Detail:
  s/MemInfo - Added an implementation of ROL's OS_Memory 24 call. Unlike the old OS_ValidateAddress call, this call should successfully report the presence of all memory areas known to the kernel. It should also correctly indicate which parts of a sparse DA are mapped in, unlike the old OS_ValidateAddress implementation.
  s/ChangeDyn - Update dynamic area handling to construct a lookup table for mapping logical addresses to dynamic areas; this is used by OS_Memory 24 to quickly locate which DA(s) hit a given region
  s/AMBControl/main - Make sure lazy task swapping is marked as disabled when AMB_LazyMapIn is {FALSE} - required so that OS_Memory 24 will give application space the correct flags
  s/ArthurSWIs - Switch OS_ValidateAddress over to using OS_Memory 24, as per ROL. For compatibility, Service_ValidateAddress is still issued for any areas which the kernel doesn't recognise (currently, OS_Memory 24 doesn't issue any service calls itself)
  s/Convrsions - ADR -> ADRL to keep things happy
  s/HAL - Fix L2PT page allocation and RAM clear to release the physical access region once they're done with it
  s/Kernel - Make the error dispatcher validate the error handler code ptr & error buffer using OS_Memory 24 before attempting to use them. If they look bad, reset to default. Should prevent getting stuck in an infinite abort loop in some situations (e.g. as was the case with ticket 279). The system might not fully recover, but it's better than a hard crash.
  s/Middle - Rework data/prefetch/etc. abort handlers so that DumpyTheRegisters can validate the exception dump area via OS_Memory 24 before anything gets written to it. Should also help to prevent some infinite abort loops. Strip 26bit/pre-HAL code to make things a bit more readable.
  hdr/KernelWS - Update comment
Admin:
  Tested on BB-xM, Raspberry Pi


Version 5.35, 4.79.2.222. Tagged as 'Kernel-5_35-4_79_2_222'

Detail:
  s/MemInfo - Fix OS_Memory 0 physical-to-logical conversion returning bad addresses for most cases due to R5 being modified by physical_to_ppn (bug introduced in revision 4.4.2.20)
  s/vdu/vdudriver - Fix some VDU driver variables not being initialised correctly when switching GraphicsV driver
  s/vdu/vdugrafv - Fix handling of VSync events from GraphicsV drivers other than driver zero
  s/vdu/vduswis - Fix abort when OS_ScreenMode 11 is passed a bad driver number
Admin:
  Tested on Raspberry Pi
  There still seems to be a bug lurking somewhere when switching to a GraphicsV driver that uses DA2; the system will crash horribly unless DA2 is already a suitable size for the initial mode change


Version 5.35, 4.79.2.206. Tagged as 'Kernel-5_35-4_79_2_206'

Detail:
  Briefly, this set of changes:
  * Adjusts PhysRamTable so that it retains the flags passed in by the HAL from OS_AddRAM (by storing them in the lower 12 bits of the size field)
  * Sorts the non-VRAM entries of PhysRamTable by speed and DMA capability, to ensure optimal memory allocation during OS startup.
  * Adjust the initial memory allocation logic to allow the cursor/sound chunk and HAL noncacheable workspace to come from DMA capable memory
  * Extends OS_Memory 12 to accept a 'must be DMA capable' flag in bit 8 of R0. This is the same as available in ROL's OS.
  * Extends OS_DynamicArea 0 to allow the creation of dynamic areas that automatically allocate from DMA capable memory. In ROL's OS this was done by setting bit 12 of R4, but we're using bits 12-14 for specifying the cache policy, so instead bit 15 is used.
  * Fixes OS_ReadSysInfo 6 to return the correct DevicesEnd value now that the IRQ/device limit is computed at runtime
  File changes:
  * hdr/OSEntries - Add definitions of the various flags passed to OS_AddRAM by the HAL. Add a new flag, NoDMA, for memory which can't be used for DMA.
  * hdr/KernelWS - Tidy PhysRamTable definition a bit by removing all the DRAM bank definitions except the first - this makes it easier to search for code which is interacting with the table. Remove VRAMFlags, it's redundant now that the flags are kept in the table. Add DMA allocation info to InitWs.
  * s/AMBControl/memmap - Updated to mask out the flags from PhysRamTable when reading RAM block sizes.
  * s/ARM600 - Strip out a lot of IOMD specific pre-HAL code.
  * s/ChangeDyn - Updated to cope with the flags stored in PhysRamTable. Implement support for DMA-capable dynamic areas. Rewrite InitDynamicAreas to insert pages into the free pool in the right order so that the fastest memory will be taken from it first.
  * s/GetAll, s/Middle - Fix OS_ReadSysInfo 6 to return the correct HAL-specific DevicesEnd value
  * s/HAL - Significant rework of initial RAM allocation code to allow the kernel workspace to come from the fastest DMA incapable RAM, while also allowing allocation of DMA capable memory for HAL NCNB workspace & kernel cursor/sound chunks. ClearPhysRAM rewritten as part of this.
  * s/MemInfo - Updated to cope with the flags stored in PhysRamTable. Add support for the new OS_Memory 12 flag. Update OS_Memory 7 to not assume PhysRamTable entries are sorted in address order, and rip out the old pre-HAL IOMD implementation.
  * s/NewReset - Remove GetPagesFromFreePool option, assume TRUE (as this has been the case for the past 10+ years). Revise a few comments and strip dead code. Update to cope with PhysRamTable flags.
  * s/VMSAv6 - Remove a couple of unused definitions
  * s/vdu/vdudriver - Update to cope with PhysRamTable flags
Admin:
  Tested in Kinetic RiscPC ROM softload, Iyonix softload, & OMAP3


Version 5.35, 4.79.2.186. Tagged as 'Kernel-5_35-4_79_2_186'

Detail:
  Revisions Kernel-5_35-4_79_2_153 and Kernel-5_35-4_79_2_161 merged with one
  trivial conflict.
Admin:
  Confirmed that this builds, but not tested on hardware.

Version 5.35, 4.79.2.147.2.15. Tagged as 'Kernel-5_35-4_79_2_147_2_15'

Previously HAL_PhysInfo took one of two routes to get the arrangment table of arrangement table size.
Reorganised to accept a subreason to allow other physical info to be requested.


Version 5.35, 4.79.2.161. Tagged as 'Kernel-5_35-4_79_2_161'

The only fake result now is the hard ROM amount, which is hardwired to 4MB and might not be correct.
Unrelated changes
 hdr.HALDevice: Assign a device for VIDC20.
 hdr.KernelWS: Reorder into ascending order, remove legacy addresses.
 s.ARM600: Move PhysSpaceSize inside :LNOT:HAL switch.
 s.Kernel: Move PhysSpaceSize inside :LNOT:HAL switch.

Version 5.35, 4.79.2.153. Tagged as 'Kernel-5_35-4_79_2_153'

Detail:
  A whole mass of changes to add high processor vectors + zero page relocation support to the Cortex branch of the kernel
  At the moment the code can only cope with two ZeroPage locations, &0 and &FFFF0000. But with a bit more tweaking those restrictions can probably be lifted, allowing ZeroPage to be hidden at almost any address (assuming it's fixed at compile time). If I've done my job right, these restrictions should all be enforced by asserts.
  There's a new option, HiProcVecs, in hdr/Options to control whether high processor vectors are used. When enabling it and building a ROM, remember:
  * FPEmulator needs to be built with the FPEAnchor=High option specified in the components file (not FPEAnchorType=High as my FPEmulator commit comments suggested)
  * ShareFS needs unplugging/removing since it can't cope with it yet
  * Iyonix users will need to use the latest ROOL boot sequence, to ensure the softloaded modules are compatible (OMAP, etc. don't really softload much so they're OK with older sequences)
  * However VProtect also needs patching to fix a nasty bug there - http://www.riscosopen.org/tracker/tickets/294
  The only other notable thing I can think of is that the ProcessTransfer code in s/ARM600 & s/VMSAv6 is disabled if high processor vectors are in use (it's fairly safe to say that code is obsolete in HAL builds anyway?)
  Fun challenge for my successor: Try setting ZeroPage to &FFFF00FF (or similar) so its value can be loaded with MVN instead of LDR. Then use positive/negative address offsets to access the contents.
  File changes:
  - hdr/ARMops - Modified ARMop macro to take the ZeroPage pointer as a parameter instead of 'zero'
  - hdr/Copro15ops - Corrected $quick handling in myISB macro
  - hdr/Options - Added ideal setting for us to use for HiProcVecs
  - s/AMBControl/allocate, s/AMBControl/growp, s/AMBControl/mapslot, s/AMBControl/memmap, s/AMBControl/service, s/AMBControl/shrinkp, s/Arthur2, s/Arthur3, s/ArthurSWIs, s/ChangeDyn, s/ExtraSWIs, s/HAL, s/HeapMan, s/Kernel, s/MemInfo, s/Middle, s/ModHand, s/MoreSWIs, s/MsgCode, s/NewIRQs, s/NewReset, s/Oscli, s/PMF/buffer, s/PMF/IIC, s/PMF/i2cutils, s/PMF/key, s/PMF/mouse, s/PMF/osbyte, s/PMF/oseven, s/PMF/osinit, s/PMF/osword, s/PMF/oswrch, s/SWINaming, s/Super1, s/SysComms, s/TickEvents, s/Utility, s/vdu/vdu23, s/vdu/vdudriver, s/vdu/vdugrafl, s/vdu/vdugrafv, s/vdu/vdupalxx, s/vdu/vdupointer, s/vdu/vduswis, s/vdu/vduwrch - Lots of updates to deal with zero page relocation
  - s/ARM600 - UseProcessTransfer option. Zero page relocation support. Deleted pre-HAL ClearPhysRAM code to tidy the file up a bit.
  - s/ARMops - Zero page relocation support. Set CPUFlag_HiProcVecs when high vectors are in use.
  - s/KbdResPC - Disable compilation of dead code
  - s/VMSAv6 - UseProcessTransfer option. Zero page relocation support.
Admin:
  Tested with OMAP & Iyonix ROM softloads, both with high & low zero page.
  High zero page hasn't had extensive testing, but boot sequence + ROM apps seem to work.


Version 5.35, 4.79.2.98.2.48. Tagged as 'Kernel-5_35-4_79_2_98_2_48'

Detail:
  Makefile - Now uses ${PERL} for running perl
  s/Kernel - Now uses correct "Bad OS_PlatformFeatures reason code" error number
  s/MemInfo - Updated list of OS_Memory 9 controllers
Admin:
  OMAP3 ROM compiles OK; untested at runtime


Version 5.35, 4.79.2.98.2.43. Tagged as 'Kernel-5_35-4_79_2_98_2_43'

Detail:
  s/MemInfo - List of OS_Memory 9 controllers now updated to include details of the ones that ROL are using, along with which numbers should/shouldn't be safe for us to expand into in the future.
Admin:
  Tested in ROM softload on RiscPC


Version 5.35, 4.79.2.114. Tagged as 'Kernel-5_35-4_79_2_114'

Detail:
  Noted reservation of IO controller type  passed to OS_Memory 9 used when
  system is running as a coprocessor. Not used by current code but we need to
  make sure that any future reservations use different numbers.
Admin:
  No functional change. Brought to our attention by Rob Sprowson.

Version 5.35, 4.79.2.103. Tagged as 'Kernel-5_35-4_79_2_103'

Add support for Cortex cache type. Extend ARM_Analyse to, where appropriate, use CPU feature registers to identify CPU capabilities.

Detail:
  s/ARMops - Support for Cortex multi-level cache (CT_ctype_WB_CR7_Lx). New ARM_Analyse_Fancy to identify CPU capabilities using feature registers.
  s/HAL - Modify pre-ARMop cache code to handle Cortex-syle caches.
  s/MemInfo - Replace ARM_flush_TLB macro call with appropriate ARMop to provide Cortex compatability
  hdr/ARMops - Update list of ARM architectures
  hdr/CoPro15ops - Deprecate ARM_flush_* macros for HAL kernels, as they are no longer capable of flushing all cache types. ARMops should be used instead.
  hdr/KernelWS - Add storage space for multi-level cache properties required for new cache cleaning code.
Admin:
  Tested under qemu-omap3. Still unable to verify on real hardware due to lack of appropriate MMU code. However new OMAP3 HAL code that uses similar cache management functions appears to work fine on real hardware.


Version 5.35, 4.79.2.98.2.2. Tagged as 'Kernel-5_35-4_79_2_98_2_2'

Added "fast" flash tool for Customer L, allowing ROMs to be sent serially at
115200 baud not 9600 baud.
Fix to VDU despatch for ARMv4 and later.
Fixes to power on delete keyboard and keyboard timeout
Implemented MemoryReadPhys and MemoryAmounts with the HAL.

Version 5.35, 4.79.2.59. Tagged as 'Kernel-5_35-4_79_2_59'

*Configure ANYTHINGsize was broken due to not setting R0 to ReadUnsigned
IIC ack message uninternationalised
OS_Memory was saying we only had 4M of RAM
VDU4 scrolling when output was switched to sprite was causing corruption
on use of CTRL-J and CTRL-K
Default SystemSize CMOS set to 32k

Version 5.35, 4.79.2.55. Tagged as 'Kernel-5_35-4_79_2_55'

Detail:
  Lots of changes since last version, at least the following:
  * Updated OS timestamp, removed alpha status
  * Negative INKEY OS version changed to &AA
  * GraphicsV is now alocated vector number &2A
  * ROM moved up to &FC000000
  * Max application slot increased to 512 Mbytes (for now)
  * Max size of RMA increased to 256 Mbytes
  * RMA is now first-created dynamic area (so it gets lowest address after
    top of application slot)
  * OS_Memory 10 reimplemeted
  * New OS_ReadSysInfo 6 values 18-22 added
  * OS_ReadSysInfo 8 gains flag bit to indicate soft power-off
  * Misc internal top-bit-set-address fixes
  * *ChangeDynamicArea can take sizes in megabytes or gigabytes
  * Magic word "&off" in R0 passed to OS_Reset powers down if possible
  * Added acceleration: block copy; CLS; text window scroll up; rectangle
    fill
  * Disabled LED flashing in page mode (liable to crash)
  * Masked sprite plot and VDU 5 text avoids reading the screen if poss...

Version 5.35, 4.79.2.48. Tagged as 'Kernel-5_35-4_79_2_48'

Ursula. Quite a hairy code merge really, so let's hope it is
worth it to someone. What you get (back after 2 or 3 years):
- much more efficient for largish numbers of DAs (relevance
  to current build = approx 0)
- fancy reason codes to support fast update of
  Switcher bar display (relevance = 0)
- support for clamped maximum area sizes, to avoid address
  space exhaustion with big memory (relevance = 0)
- better implementation of shrinkable DAs, performance
  wise (if lots of DAs, relevance = approx 0)
- support for 'Sparse' DAs. Holey dynamic areas, Batman!
  (relevance, go on someone use the darned things)
Moderately development tested on HAL/32bit ARM9 desktop.
Note the Switcher should be compiled to use the new
reason codes 6&7, for fabled desktop builds.

Also, during this work, so I could see the wood for the
trees, redid some source code clean up, removing pre-Medusa
stuff (like I did about 3 years ago on Ursula, sigh). That's
why loads of source files have changed. The new DA stuff
is confined pretty much to hdr.KernelWS and s.ChangeDyn.

Ta.

Version 5.35, 4.79.2.38. Tagged as 'Kernel-5_35-4_79_2_38'

  hooks to give correct mapping info for OS_Memory 0
  same for OS_ReadMemMapEntries
  same for OS_FindMemMapEntries
  Lazy fixup routine no longer assumes an abort in current
  app space must be a truant page. However, work in this
  area not complete (no support yet for abort handler code
  in app space itself, eg. for C trampoline)
Good to know this will be a big performance boost when
our products use one monolithic application (sarcasm).
Ta

Version 5.35, 4.79.2.32. Tagged as 'Kernel-5_35-4_79_2_32'

Reimplement Lazy task swapping, an amusing idea from Ursula, would have done it sooner but couldn't be bothered (humour).

Currently activates for all ARMs flagged as base-restored
abort model. No handling of eg. StrongARM pre-revT bug, but
then the kernel no longer runs on StrongARM (progress).
Still some details to fix: all aborts in current app space
assumed to be missing pages, but this must be fixed to
handle abort code in app space, things like debuggers
marking code read only.

Plus, small fixes:
  OS_Memory 8 returns vaguely useful info for RAM,VRAM
  in HAL build (temporary partial implementation)
  Broken handling of old BBC commands with (fx,tv etc)
  with no spaces fixed (fudgeulike code from Ursula,
  now 32-bit).

Version 5.35, 4.79.2.31. Tagged as 'Kernel-5_35-4_79_2_31'

* Kernel puts sensible default FIQ handler in through the HAL.
* Fix to temporary page uncaching code.

Version 5.35, 4.79.2.30. Tagged as 'Kernel-5_35-4_79_2_30'

spectacular new OS_Memory reason codes 13 map permanent I/O space, return logical address 14 access temporary physical mapping 15 release temporary physical mapping

DA creation and I/O space creation now avoid collision if address
space fills

Version 5.35, 4.79.2.28. Tagged as 'Kernel-5_35-4_79_2_28'

First attempt at ARM9 support, and general clean-up of old ARM-specific code, now using vectored ARMops.

Not tested.

Version 5.35, 4.79.2.14. Tagged as 'Kernel-5_35-4_79_2_14'

Imported from Kernel 5.36.

It says "Abort on data transfer".

* Added ARM_IMB and ARM_IMBRange SWIs as recommended by ARMv5.
* Some early prototype HAL bits popped in - a lot of source restructuring still
  to come.
* New debug target creates an AIF image with debug information, and translates
  this into an ASCII object file for the 16702B logic analyser.

Version 5.35, 4.79.2.1. Tagged as 'Kernel-5_35-4_79_2_1'

Details:
  The Kernel will now compile to produce a pure 32-bit system if No26bitCode is
  set to TRUE.
  If No26bitCode is FALSE, then the Kernel will be a standard 26-bit Kernel,
  although some internal changes have taken place to minimise compile
  switches between the two cases. See Docs.32bit for more technical info.

  The hardest part was the flood-fill...

Other changes:
  Pointer shape changes now take place on the next VSync, rather than actually
  WAITING for the VSync. Turning the Hourglass on shouldn't slow your machine
  down by 5% now :)

  Lots of really crusty pre-IOMD code removed.

Admin:
  Tested in 32 and 26-bit forms in a limited desktop build. Basically, this
  will need to see a lot of use to iron out difficulties. I'd like anyone who
  has a non-frozen project to at least attempt using this Kernel.

Version 5.23. Tagged as 'Kernel-5_23'

Video memory now limited to 8M (instead of 16M) to make room.
OS_Memory 7 now reports ROM correctly when the image is >4M.

Version 4.76. Tagged as 'Kernel-4_76'