Gregory Nutt
|
a9864bc100
|
Extend virtual/physical address conversions to include addresses in shared memory.
|
2014-09-23 16:03:08 -06:00 |
|
Gregory Nutt
|
cce614c67e
|
Fix some inconsistent field name in struct task_group_s: addrenv should be tg_addrenv.
|
2014-09-23 16:01:44 -06:00 |
|
Gregory Nutt
|
df26163f1f
|
Add logic necessary to handler remapping of shared memory on contex switches
|
2014-09-23 13:19:30 -06:00 |
|
Gregory Nutt
|
e94adf8ff4
|
Ooops... a file that I forgot to add yesterday
|
2014-09-17 09:52:07 -06:00 |
|
Gregory Nutt
|
04f75c42cc
|
Add a sharable version of arm_virtpgaddr()
|
2014-09-16 16:49:44 -06:00 |
|
Gregory Nutt
|
d936f1a918
|
First round of changes to get the ELF configuration building again
|
2014-09-16 15:37:05 -06:00 |
|
Gregory Nutt
|
c20907b86c
|
These files were deleted and moved to a different location (see previos commit)
|
2014-09-16 13:36:51 -06:00 |
|
Gregory Nutt
|
8058de7874
|
remove tailing blank line
|
2014-09-16 13:36:14 -06:00 |
|
Gregory Nutt
|
bd6f7c50e8
|
Move common/up_signal_dispatch.c to armv6-m, armv7-m, and armv7-a. The armv7-a version needs to be different to handle the case where we are dispatch kernel mode signals when running under a user mode group
|
2014-09-16 13:35:29 -06:00 |
|
Gregory Nutt
|
8c3fd5a162
|
Correct stack handling is signal deliver to user processes
|
2014-09-16 13:33:13 -06:00 |
|
Gregory Nutt
|
ebc10e0237
|
Move some share-able logic from sama5/sam_pgalloc.c to armv7-a/arm_physpgaddr.c
|
2014-09-16 13:31:48 -06:00 |
|
Gregory Nutt
|
4a04ea48e4
|
Move some share-able logic from sama5/sam_pgalloc.c to armv7-a/arm_physpgaddr.c
|
2014-09-16 13:31:24 -06:00 |
|
Gregory Nutt
|
9879f47bbb
|
Move some share-able logic from sama5/sam_pgalloc.c to armv7-a/arm_physpgaddr.c
|
2014-09-16 13:29:43 -06:00 |
|
Gregory Nutt
|
4f1ca4cbb7
|
Partial fix to an cache cleaning problem
|
2014-09-15 16:04:42 -06:00 |
|
Gregory Nutt
|
35a282bc30
|
Fix an error in a debug statement
|
2014-09-15 15:15:57 -06:00 |
|
Gregory Nutt
|
e2952e4088
|
Eliminate a bug introduced in last checking
|
2014-09-15 15:10:50 -06:00 |
|
Gregory Nutt
|
d14d3b29bb
|
ARMv7-A: Improvements to assertion output for kernel mode
|
2014-09-15 15:03:55 -06:00 |
|
Gregory Nutt
|
9aae0adffa
|
If we are configured to use a kernel stack while in SYSCALL handling, then we need to switch back to the user stack to deliver a signal
|
2014-09-15 11:38:48 -06:00 |
|
Gregory Nutt
|
949cc1b20c
|
Fix a typo in system call when fetching parameter from the stack: regs[REG_PC]+4 is the address, not regs[REG_PC+4]
|
2014-09-15 10:55:10 -06:00 |
|
Gregory Nutt
|
946b916f69
|
Initial integration of kernel stack (does not work)
|
2014-09-14 11:19:34 -06:00 |
|
Gregory Nutt
|
16ddffc941
|
Add the initial implementation of the process kernel stack logic. Not yet integrated into the main OS logic nor tested.
|
2014-09-14 09:53:54 -06:00 |
|
Gregory Nutt
|
ffff51c1b1
|
Rename everything associated with the dynamic process stack to ustack to make room in the name space for a kstack
|
2014-09-14 09:10:09 -06:00 |
|
Gregory Nutt
|
d956936535
|
Add logic need to manage a virtualized stack. Not yet incorporated into base OS logic.
|
2014-09-13 13:45:35 -06:00 |
|
Gregory Nutt
|
5d4f336b1f
|
Move static helper routines from arm_addrenv.c and may them global so that they can be shared both forthcoming stack address environment logic.
|
2014-09-13 13:17:44 -06:00 |
|
Gregory Nutt
|
f4bcb27962
|
Add a configuration option for dynamic stack management
|
2014-09-13 12:25:32 -06:00 |
|
Gregory Nutt
|
fe48417a96
|
Comsetic updates to comments, debug output
|
2014-09-12 10:31:58 -06:00 |
|
Gregory Nutt
|
fba35f0c41
|
ARMv7-A: Modify up_fullcontextrestore() for CONFIG_BUILD_KERNEL. It changed CPSR while in kernel. That will crash is the new CPSR is user mode while executing in kernel space. Fixed by adding a SYS_context_restore system call. There is an alternative, simpler modification to up_fullcontextrestore() that could have been done: It might have been possible to use the SPSR instead of the CPRSR and then do an exception return from up_fullcontextrestore(). That would be more efficient, but I never tried it.
|
2014-09-12 08:04:27 -06:00 |
|
Gregory Nutt
|
6084fad7e0
|
Fix logic for returning from exceptions to user-mode contexts
|
2014-09-11 18:43:30 -06:00 |
|
Gregory Nutt
|
62e608be8c
|
All tasks, even user mode tasks, must start in supervisor mode until they get past the start-up trampoline
|
2014-09-11 18:42:52 -06:00 |
|
Gregory Nutt
|
4821587509
|
Update some comments/function headers
|
2014-09-11 17:15:26 -06:00 |
|
Gregory Nutt
|
401b8543cd
|
Tighten up some ARM assembly language. You can always do better
|
2014-09-11 15:12:08 -06:00 |
|
Gregory Nutt
|
bd7f16d16a
|
ARMv7-A: Exception register save/restore needs to work a little differently if we support user mode processes
|
2014-09-11 14:34:10 -06:00 |
|
Gregory Nutt
|
998c5ee0a9
|
Trivial kernel build related fixes for consistency
|
2014-09-11 12:35:23 -06:00 |
|
Gregory Nutt
|
1f23ad9bad
|
Misc fixes to repair some of the breakage to the SAMA5D4-EK elf configuration caused by changes for the knsh configuration
|
2014-09-11 10:31:12 -06:00 |
|
Gregory Nutt
|
006cf7d745
|
Add logic to initialize the per-process user heap when each user process is started
|
2014-09-10 15:55:36 -06:00 |
|
Gregory Nutt
|
2da0392ae2
|
SAMA5D4-EK: These configurations now use the fixed DRAM mapping for manipulating the page memory pool.
|
2014-09-10 08:44:09 -06:00 |
|
Gregory Nutt
|
6238e87aaa
|
Add configuration to use the fixed DRAM mapping for the page pool (if available) instead of remapping dynamically to access L2 page tables and page data. Also, add logic in address environment creation to initialize the shared data at the beginning of the .bss/.data process memory region.
|
2014-09-10 08:41:01 -06:00 |
|
Gregory Nutt
|
8a99c421ff
|
pcDuino: Several fixes so that it still builds after other Cortex-A changes.
|
2014-09-10 06:24:39 -06:00 |
|
Gregory Nutt
|
aaf190dcf6
|
ELF relocations. Some relocation types do not have a named symbol associated with them. The design did not account for that case
|
2014-09-09 16:52:51 -06:00 |
|
Gregory Nutt
|
8b64dc003e
|
SAMA5D4-EK: In kernel build with address environment, need logic to map user virtual addresses to physical addresses, and vice versa
|
2014-09-07 19:25:30 -06:00 |
|
Gregory Nutt
|
53bd807186
|
Fix loop counter... was overrunning a table on larger ELF files
|
2014-09-07 14:42:04 -06:00 |
|
Gregory Nutt
|
dcc711f3f2
|
Correct size comparison (pages vs. sections)
|
2014-09-07 13:47:01 -06:00 |
|
Gregory Nutt
|
3dd3b1f5e2
|
The 'make export' target needs to bundle up the user C startup file (crt0), not the kernel head object
|
2014-09-04 13:31:34 -06:00 |
|
Gregory Nutt
|
70e5350942
|
Mostly cosmetic changes
|
2014-09-04 10:28:38 -06:00 |
|
Gregory Nutt
|
15e439d6de
|
I love/hate conditional compilation
|
2014-09-03 11:43:23 -06:00 |
|
Gregory Nutt
|
12775801c9
|
Add support for delivery of use-mode signals in the kernel build.
|
2014-09-02 15:58:14 -06:00 |
|
Gregory Nutt
|
8557f1a1bb
|
Space at the beginning of the process data space is now reserved for user heap management structures. In the kernel build mode, these heap structures are shared between the kernel and use code in order to allocate user-specific data.
|
2014-09-02 11:21:23 -06:00 |
|
Gregory Nutt
|
4d8367a009
|
sbrk() need to initialized the memory manager on the first call
|
2014-09-02 08:05:11 -06:00 |
|
Gregory Nutt
|
587520a7d2
|
Completes the implementation of sbrk() (untested)
|
2014-09-01 10:46:51 -06:00 |
|
Gregory Nutt
|
4537a905f6
|
ARMv7 address environment: Static functions not marked static
|
2014-09-01 08:49:08 -06:00 |
|