- 14 Apr, 2018 1 commit
-
-
Jeffrey Lee authored
Detail: kernel/s/k_body - Update AbortFindHandler to check that SP & SL are both word aligned, especially because the code will attempt to load from SL to check for the stack chunk magic marker. Admin: Tested on BB-xM Avoids some "abort while in abort handler" type scenarios if an abort occurs while SP/SL are invalid Version 5.95. Tagged as 'RISC_OSLib-5_95'
-
- 29 Oct, 2017 1 commit
-
-
Robert Sprowson authored
The ctl part of rlib has various memory leaks (ref https://www.riscosopen.org/forum/forums/4/topics/9503) reflecting its abandoned state. See the FrontEnd module for a more flexible implementation of the same GUI-in-a-text-file scripting language. The ctl API was not in C release 3 (it post dates that) nor in C release 4 manual, and the whole of rlib was removed from C release 5 in favour of the Toolbox. The change log at the top of rlib/s/rl_spare shows it was briefly added in early 1990 then removed in April 1991. 26 years later, cull it from the Makefile too. Version 5.94. Tagged as 'RISC_OSLib-5_94'
-
- 03 Sep, 2017 1 commit
-
-
Jeffrey Lee authored
Detail: kernel/s/k_body - Eliminate a couple of global OS_SynchroniseCodeAreas calls in _kernel_init and _kernel_system s/initmodule - Eliminate a global OS_SynchroniseCodeAreas call in _Shared_Lib_Module_SWI_Code Admin: Tested on iMx6 Improves performance with SMP ROMs, where global D-cache clean isn't really possible Version 5.93. Tagged as 'RISC_OSLib-5_93'
-
- 28 May, 2016 1 commit
-
-
Jeffrey Lee authored
Detail: kernel/s/k_body, s/initmodule - Remove NOPs from after PSR manipulation macros; the macros now fully take care of ARM2 & StrongARM compatibility for us Admin: Tested PlingSystem build on (ARM3) RISC OS 3.1 Requires HdrSrc-2_63 for ARM2 compatibility Version 5.92. Tagged as 'RISC_OSLib-5_92'
-
- 25 May, 2016 1 commit
-
-
Jeffrey Lee authored
Detail: s/longlong - Copy-paste error led to conditional being on wrong line Admin: Tested with Raspberry Pi ROM build Version 5.91. Retagged as 'RISC_OSLib-5_91'
-
- 24 May, 2016 1 commit
-
-
Jeffrey Lee authored
When multiple CPU architectures are supported, allow dynamically linked clients to be given optimal versions of routines where possible Detail: This set of changes adds support for representing architecture-specific variants of routines within the export tables which the shared C library uses to fill in the client's stubs. This allows builds of the module which target multiple architectures (e.g. IOMD, Raspberry Pi, or softloads) to offer the most optimal versions of routines to clients wherever multiple variants exist. If only one architecture is supported, only one version of each routine will be built, and no variant tables will be generated. Currently routine selection for staticly linked clients isn't supported - staticly linked targets will only use the most backwards-compatible version of the routines, as before. Also not all routines are catered for yet (e.g. _ll_udiv) File changes: - s/h_common - New file containing macros used during variant table construction - Makefile - Set SHARED_C_LIBRARY to {TRUE} for SCL builds of assembler code, as per the SHARED_C_LIBRARY #define - kernel/s/k_body, s/longlong - For routines which have multiple variants available for the target architectures, build all variants, and generate variant table entries using the new RoutineVariant macro. This has required some reordering of the different variants to make sure that when both versions are built, the most backwards-compatible one will be first (and thus will be called by staticly linked clients) - s/h_modmacro - "Entry" macro (when building SCL export tables) changed to use DCD instead of B, to allow byte-aligned symbols to be represented (low bits of symbol address is now used as 'is a variant table' flag) - s/initmodule - Client stub initialisation rewritten to take into account the B -> DCD change, and to detect and evaluate variant entries - clib/s/cl_entries, kernel/s/k_entries, kernel/s/k_entries2 - Update library entry tables to reference the variant forms of routines where relevant Admin: Tested on Raspberry Pi 1, 2, 3 (ROM), RISC OS 3.1 & StrongARM 3.7 (softload 26bit build) Requires Kernel-5_35-4_79_2_325 Version 5.91. Tagged as 'RISC_OSLib-5_91'
-
- 14 May, 2016 1 commit
-
-
ROOL authored
Detail: When in the default C locale localtime() would return the wrong time for nonzero timezones. Admin: Tested with softloaded C library for each ROM release supported. Submission from Rick Murray. Version 5.90. Tagged as 'RISC_OSLib-5_90'
-
- 08 May, 2016 1 commit
-
-
Jeffrey Lee authored
Prefer CPS over MSR for PSR manipulation. Use UDIV/SDIV for general-purpose division, and UMULL for /10. Detail: s/k_body: - _kernel_irqs_off and _kernel_irqs_on now use CPS for disabling/enabling IRQs as opposed to MSR. Apart from being shorter code sequences, it's generally a faster instruction. - __rt_udiv and __rt_sdiv (and aliases) now use the UDIV and SDIV instructions if building for ARMv7VE - __rt_udiv10 and __rt_sdiv10 (and aliases) now use UMULL to multiply by 1/10 when building for targets with long multiply support, as this is faster than the old method. UDIV/SDIV can be fast too, but only for small numbers, making UMULL the best for the general case. Admin: Tested on Cortex-A15 Prototyping of division routines on assorted CPUs shows that UDIV/SDIV is generally between 20% and 400% faster than the old routine (Cortex-A7, Cortex-A53), or up to 1300% faster on Cortex-A15 (the CPU does not like the old routine!) Division by 10 is now about 20% faster across all appropriate CPUs Version 5.89. Tagged as 'RISC_OSLib-5_89'
-
- 01 Mar, 2016 2 commits
-
-
Ben Avison authored
Detail: AcquireMutex doesn't have the same functionality as Arm2Swp, best not to make people think they might do. Also add an extra literal pool because otherwise some instructions don't reach when doing APCS-32 softload builds. Retagged as 'RISC_OSLib-5_88'
-
Robert Sprowson authored
Retagged as RISC_OSLib-5_88.
-
- 29 Feb, 2016 1 commit
-
-
Ben Avison authored
Detail: The SWP and SWPB instructions have finally been removed in ARMv8, after having been deprecated for a very long time. This version adds alternative versions of code that used to use them with ones that use LDREX/STREX and LDREXB/STREXB instead. Soft-loadable C libraries will choose between implementations at runtime using OS_PlatformFeatures; ROM builds only include the appropriate version for the target hardware. Admin: Tested on Raspberry Pi 3. Version 5.88. Tagged as 'RISC_OSLib-5_88'
-
- 26 Feb, 2016 1 commit
-
-
Robert Sprowson authored
C version of the definition from Hdr:Messages. Retagged as RISC_OSLib-5_87.
-
- 13 Feb, 2016 1 commit
-
-
ROOL authored
Detail: When RISC_OSLib is used to generate menus using menu_new and menu_extend there are two (undocumented?) limits on the total menu size: - 64 items on a menu. Any more are silently ignored. - 1024 bytes of indirect text. Above this, any items have several bytes of random rubbish displayed in the menu entry. rlib.c.menu (r4.6) has been changed to: - remove menu_workarea which was on the stack, and caused limits. - only use malloc'd storage areas for the menu. - reduce the data copied by at least 50% by changing processing from: copy; update; copy to : realloc extend (& copy?); update; realloc shrink (no copy). - remove the limits by adding more storage dynamically. - simplify code, which reduces the object code size by about 7%. Admin: The changes have been tested on an Iyonix running RO5.23, and VRPC running RO4.39, with a large C program with about 80 menus. The only noticeable effects were beneficial. It has also been tested with artificially low ADDITEMS and ADDBYTES. Submission from Martin Avison. Version 5.87. Tagged as 'RISC_OSLib-5_87'
-
- 07 Feb, 2016 2 commits
-
-
Robert Sprowson authored
Function dboxtcol() always opens colour pickers as transient submenus (in the flags to ColourPicker_OpenDialogue, plus listening for the menus deleted message) but chose to open the dialogue at the current mouse pointer, so obscuring part of the menu rather than opening it where the Style Guide wants it. wimp.h: Add a struct definition of menuwarn messages. txtedit.c/dbox.c/ctl.c: Use the menuwarn struct rather than indexing the words[]. dboxtcol.c: Provided the last wimp event was the menu warning, use the x/y coordinates from that instead of the mouse pointer. Remove some local defines in favour of central ones. Make sure we use 0x80000000/0x7F000000 like the PRM says for the other 2 corners of the dialogue, not 0,0. Tested in a disc build of !Draw. Version 5.86. Not tagged
-
Robert Sprowson authored
-
- 30 Jan, 2016 1 commit
-
-
ROOL authored
Admin: Part of submission from Martin Avison. Not tagged.
-
- 08 Nov, 2015 1 commit
-
-
ROOL authored
Detail: Export Hdr:ATA if present. Admin: Tagged as RISC_OSLib-5_86-1 since existing binaries unchanged.
-
- 08 Aug, 2015 1 commit
-
-
Robert Sprowson authored
See ISO9899:1999 7.25.3.2.1. Retagged as RISC_OSLib-5_86.
-
- 01 Aug, 2015 1 commit
-
-
Robert Sprowson authored
Retagged as RISC_OSLib-5_86
-
- 30 Jun, 2015 1 commit
-
-
Jeffrey Lee authored
Detail: c/armsys - _kernel_language will return NULL if the PC was outside any known library language block, so make sure to check for that before dereferencing the pointer Admin: Builds, untested Version 5.86. Tagged as 'RISC_OSLib-5_86'
-
- 15 May, 2015 1 commit
-
-
Robert Sprowson authored
Not tagged.
-
- 04 Apr, 2015 1 commit
-
-
Robert Sprowson authored
Make sure 'Edit' isn't left shaded by a previous TaskWindow opening. Retagged as RISC_OSLib-5_85.
-
- 03 Apr, 2015 3 commits
-
-
Robert Sprowson authored
Trim any #if FALSE switches over a decade old, and keep #if TRUE. Accept EDIT2 and FMTTEXT1 bracketed code. Remove unused internal functions txt1_dooaction txt1_dosetoaction. Retagged as RISC_OSLib-5_85.
-
Robert Sprowson authored
The txt library part of RISC_OSLib was essentially contrary to everything Acorn was telling its developers, to use the global clipboard. Additionally, support is added for swap case within a selection. txt.h: Add a new charoption type 'txt_READONLY' to inform txt that the buffer is notionally read only, this is used to supress 'Paste' in the selection menu when appropriate. Correct some typos in the comments. txt1.h: Kill off modula 2 dummy structure member. txtar.h: Remove unused function export. txtfile.h: Rename basicimport to be consistent with the text version of the same function. We use 'import' to denote RAM transfers and 'insert' to denote file/scrap transfers throughout. txtmisc.h: Redundant internal functions removed. txtundo.h: Add a new undo operation type 't' for swap case, to avoid polluting the undo buffer with an entire copy of the text where only the case changed. Kill off modula 2 double pointer requirement. txtedit.h: Unused structure member 'selectctl' removed. txt.c: Kill off modula 2 double pointer requirement. txtar.c: Refactor message despatch with a switch statement so the save/load/open operations are explicitly checked for, in case the application enables other messages. txtedit.c: Adopt MOVERWRITE, been enabled since 1988 so is probably good to keep. Dynamically generate the Select menu by first sending a clipboard request, and fading Paste if no reply comes back. Implement changed mouse selection logic. Implement different hotkeys and caret navigation. Implement copy and paste/replace operations. Implement swap case operation. txtfile.c: Type corrections and function rename admin. Implement copy and paste/replace operation for detokenised BASIC. txtmisc.c: Supporting functions for clipboard added, supporting functions for ArcEdit removed. txtscrap.c: Make sure the caret is visible for programmatical zero sized selections. txtundo.c: Allow suspension of undo during known complex operations, such as import via RAM transmit. This was a longstanding bug where the import buffer was grabbed in ~4k chunks (even if only 1 byte was being transferred) which in turn resulted in a +4000 undo insertion and a -3999 removal, which given the default undo buffer is only 5k would result in it failing the reversibility test, so no undo was possible. Now, during a RAM transmit undo is suspended until the total transfer size is known, and only that data is placed in the undo buffer (subject to the same 5k reversibility limit). Add new undo type 't' for swap case operation. rlibdata.s: 4 new ints and 1 new BOOL, so RlibSpace increases by 5. Version 5.85. Tagged as 'RISC_OSLib-5_85'
-
Robert Sprowson authored
Add xferrecv_last_ref to get hold of the Wimp message id of the request for a scrap file data save, when the RAM transfer gets no reply. This is useful to allow the data save to be distinguished from a file being dragged to an application, without having to resort to string compares against <wimp$scrap>. SWIOptions: Also add RTC to SWI exports, and remove redundant ones. armsys.c: Treat cmd_string as a pointer. xferrecv.c/.h/rl_entries: New function xferrecv_last_ref. Version 5.84. Tagged as 'RISC_OSLib-5_84'
-
- 29 Mar, 2015 2 commits
-
-
Robert Sprowson authored
Now ObjAsm can ALIGN with arbitrary numbers, use the same trick FileCore does in its header by padding with newlines. Not tagged.
-
Robert Sprowson authored
-
- 16 Jan, 2015 2 commits
-
-
Jeffrey Lee authored
Detail: s/modulewrap - Escape some dollars contained in strings to avoid warnings from objasm Admin: Resulting binary unchanged (FilerAction) Version 5.83. Retagged as 'RISC_OSLib-5_83'
-
Jeffrey Lee authored
Detail: kernel/s/k_body, s/initmodule - Escape some dollars contained in strings to avoid warnings from objasm Admin: Resulting binary unchanged Version 5.83. Retagged as 'RISC_OSLib-5_83'
-
- 23 Aug, 2014 1 commit
-
-
Robert Sprowson authored
dbox.c: address of cancelled by dereference event.c: spelling txtar.c: don't grab the caret when clicking toggle size txtedit.c: use BOOL for a boolean win.c: when looking at the behind window handle we want -1 not 'DUD' (which happens to be -1) wimp.h: add comment about window handle -3 Retagged as RISC_OSLib-5_83. Version 5.83. Not tagged
-
- 01 Aug, 2014 2 commits
-
-
Jeffrey Lee authored
Detail: Makefile - Get the -depend option from ${STDTOOLOPTIONS}; also allows us to pull in the throwback flag. Fix all ${AS} calls to use ${AFLAGS} so that dependencies are tracked properly. Admin: Identical to previous binary Version 5.83. Retagged as 'RISC_OSLib-5_83'
-
Jeffrey Lee authored
Detail: rlib/s/rlibdata - Reduce RlibSpace by 1, to take into account the removal of a static from rlib/c/xfersend Admin: Tested on BB-xM Fixes 'Static data size in library and stub disagree' error when running ROM apps Version 5.83. Retagged as 'RISC_OSLib-5_83'
-
- 31 Jul, 2014 1 commit
-
-
Robert Sprowson authored
menu.c/menu.h: Add a new menu_setcolours() function to allow the fore/background colour bits to be set on a menu entry. colourmenu.c: Use the new menu_setcolours() function instead of making assumptions about menu's internal structure layout. dbox.c/dbox.h: New functions dbox_hidefield and dbox_unhidefield added to complement the fade/unfade ones. As dbox_findicon/findiconbefore are only ever used to find writables, rename them as such. Also, make them find writable (type 15) AND writable draggable (type 14) icons. Allow navigation by tab/shift tab through icons like the Style Guide chapter 16 says we should. In fact it's a little odd that dbox is emulating 'Ktar' validation itself manually when the Wimp has done this since version 2.92, but maintaining it is less effort than going and checking all the templates. Fix bug when looking for leading capital letter matches to complete a dbox, the previous code was looking at the cached copy of the icon flags in d->window rather than the ones the Wimp is actually showing on screen (as might have been modified by dbox_fadefield() for example), so it was possible to complete a dbox for an icon which wasn't visible or selectable. Fix bug when looking for leading capital letter matches when the icon doesn't have any text - the template file format is defined as being 0x0D terminated. This would then lead to 'R' and 'S' being accepted as matches having run off the end of the text buffer and into the validation string (which usually contains an 'R' or 'S' validation first). Changed to accept any control terminator. event.c: When force closing the current menu, don't forget to set the two state variables back to 0. fileicon.c/resspr.c/txt1.c/txtundo.c/wimp.c: Use existing pointer type definitions. flex.c: Treat bools as bools and pointers as pointers. Say something useful in an error box if flex fails its internal consistency check, rather than just data aborting. font.c: Remove ROM conditionals from around font_readdef() function, always include the function. fontselect.c: Use font_readdef() rather than local copies of the exact same function. txtedit.c/txtfile.c/txtfind.c/txtedit.h: Remove redundant 'deletepending' state variable, and confusion over 'selectrecent' being a bool. Call os_byte directly rather than passing to os_cli("*FX"). Other minor tidy ups. wimpt.c: Mark messages list as a constant. xferrecv.c: Also preserve the size member while recycling the xferrecv_ack message block, rather than corrupting it as 256. xfersend.c/xfersend.h: Take out the leaf name formulator as a common function for the two places to use it. This also allows the function to return the length of the leaf name so the correct message block size can be used. Inspection with MsgMon/!Reporter was showing that all xfersend initiated saves were passing the leaf name padded to 256 with garbage. Reindent a few functions. Make the draggable sprite replacement follow the same validation string rules as the Wimp, previously only ";s" was accepted, so upper case 'S' and validation strings that start with the sprite validation didn't work. xfersend_pipe() function prototype changed to match xfersend() more closely - not being able to respond to the context of a wimp_eventstr made this function practically unusable. akbd.h: Windows logo and menu keys defined. flex.h/help.h/wimpt.h Some typos in comments. wimp.h: Missing wimp message action codes and structure defs added for completeness. Version 5.83. Not tagged
-
- 08 Jul, 2014 1 commit
-
-
Robert Sprowson authored
Edit-1_70 and SrcEdit-1_28 no longer require these private headers. Version 5.83. Not tagged
-
- 15 Jun, 2014 1 commit
-
-
Ben Avison authored
Detail: These were previously languishing in the C compiler sources, but they belong in the C library. Note however that the library does not implement the functions described in these headers at the present time, so any attempt to use them will result in link-time errors. Admin: Untested (known not working) Version 5.83. Tagged as 'RISC_OSLib-5_83'
-
- 08 May, 2014 1 commit
-
-
Robert Sprowson authored
menu.c: if NULL return -1, this has the added advantage that a menu maker (see event_attachmenumaker) can return NULL to mean "no menu thanks" and not data abort in event.c dbox.c: if NULL return 0, this is somewhat consistent with template_syshandle() menu.h: typo in comment corrected Version 5.82. Tagged as 'RISC_OSLib-5_82'
-
- 25 Mar, 2014 1 commit
-
-
Ben Avison authored
Detail: Large file support made use of long longs in various definitions. These were switched out in PCC mode, but no allowances were made for the fact that C90 doesn't define long long either. Remember that to force cc to use non-builtin standard headers, you need to point at these using the -j switch (e.g. -jC:). The only version of cc which featured the C90-incompatible stdio.h was 5.70. Admin: Checked that the header now works when included in -pcc, -c90 and -c99 modes. Version 5.81. Tagged as 'RISC_OSLib-5_81'
-
- 21 Jan, 2014 1 commit
-
-
Robert Sprowson authored
msgs.c: includes a MessageTrans parser that duplicates MessageTrans_Lookup (needed because the lookup function returns pointers to static C strings). However, did not allow '?' wildcards in the message token names. Version 5.80. Tagged as 'RISC_OSLib-5_80'
-
- 20 Jan, 2014 1 commit
-
-
Jeffrey Lee authored
Detail: kernel/s/k_body - At CopyUpDone in _kernel_system(), a big block of conditional code was converted to unconditional as part of the 32bit conversion process. However one line for setting up the error handler remained conditional, potentially preventing the correct error buffer pointer being set when about to start the child task. Admin: Tested on BB-xM Version 5.79. Tagged as 'RISC_OSLib-5_79'
-
- 24 Oct, 2013 1 commit
-
-
Robert Sprowson authored
Fix for problem passing long command lines via system(), while there are good arguments to leave the threshold at 255, non DDEUtils aware programs get brutally truncated command strings, and since the old threshold of 255 applied when the kernel handled 256 command lines, upping the limit to 1k is no worse than before. Fix for Edit failing to draw lines of text longer than 192 characters (ticket #350). Several factors were conspiring here: on a mode change the window width in characters wasn't being clamped, whereas user entered limits in the menu were, then the internal buffer ("PAINTBUFSIZE") was too short so lines got truncated, and when larger than 256 got wrapped modulo 256 due to the use of an array of chars. Now set at compile time with BIG_WINDOW_SIZE_LIMIT with a new theoretical maximum of 8192. Currently limited to 480 since it uses stack variables. In the C library: armsys.c: adapt the threshold at which to use DDEUtils based on the kernel version number scanf.c: squash a warning time.c: mark the table of month lengths as constant so we can claw back 13 words of static workspace without having to change the stubs In RISC OS lib: bbc.c: remove local definitions of some SWI numbers template.c: squash a warning txt.c: delete Modula-2 support txt1.c: compare pointer with NULL not an integer txtar.c: introduce BIG_WINDOW_SIZE_LIMIT, rationalise a few sprintf's, mark private functions as static txtundo.c/txtedit.c: squash old style function warnings txtoptmenu.c: clamp the window width read from Edit$Options properly, delete long disabled code txtscrap.c: dead function deleted Version 5.78. Tagged as 'RISC_OSLib-5_78'
-