Don't enable AHB support in HAL on 64-bit systems, it's not 64-bit clean
and is not meant to be. Provide replacement definition for KSEG1ADDR
when it's not defined (that includes 64-bit mips). Remove other
replacement definitions, they are not needed anymore.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@4064 0192ed92-7a03-0410-a25b-9323aeb14dbd
Don't assume that both iowrite32() and iowrite32be() return values in
the _OS_REG_WRITE definition. _OS_REG_WRITE discards the value anyway.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@4061 0192ed92-7a03-0410-a25b-9323aeb14dbd
32-bit PowerPC only gets iowrite32be() and ioread32be() in version
2.6.20. This applies only to the merged "powerpc" architecture
identified by CONFIG_PPC_MERGE. The old "ppc" architecture goes away in
Linux 2.6.27, so CONFIG_PPC_MERGE is no longer valid.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@4029 0192ed92-7a03-0410-a25b-9323aeb14dbd
It's obsolete and useless now, when the HAL sources are open. It also
misuses ath_hal_debug in a way incompatible with the use of
ath_hal_debug inside the HAL.
The mmiotrace facility is available in the kernel and doesn't require
any changes to the sources.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@4020 0192ed92-7a03-0410-a25b-9323aeb14dbd
Provide forward declaration of struct ath_hal in ah_osdep.h. Implement
OS_GETUPTIME (the implementation is quite primitive, as it's only used
once for some obscure condition in 5212 chips). Implement HALDEBUG()
and ath_hal_ether_sprintf().
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@4019 0192ed92-7a03-0410-a25b-9323aeb14dbd
Check that AH_PRIVATE(_ah)->ah_eepromDetach is not NULL before calling
it. This fixes kernel oops in case if the attach function for the chip
fails before EEPROM support initialization and calls the corresponding
detach function.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3991 0192ed92-7a03-0410-a25b-9323aeb14dbd
When the compiler tries inlining static functions, it can notice that
some variables may not be initialized in the functions being inlined.
Provide initialization for output variables in all branches of
execution, no matter how anomalous. Add asserts in some cases.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3988 0192ed92-7a03-0410-a25b-9323aeb14dbd
gcc warns that "array subscript is above array bounds" with -O3 since
pvar is assigned before _i is checked.
Eliminate _i, use a new ppvar variable, which is a pointer to the
reference table. Terminate reference tables with NULL instead of using
ARRAY_SIZE. Fix types of reference tables to match those actually used
by the code.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3987 0192ed92-7a03-0410-a25b-9323aeb14dbd
Add targets for both HAL binaries without $(obj), which is needed for
Linux 2.4 build system. Keep targets with $(obj) for Linux 2.6.
Use $(HALBIN) for non-unmangled HAL and $(HALOBJ) for unmangled HAL.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3836 0192ed92-7a03-0410-a25b-9323aeb14dbd
Conversions in place should be avoided in the build system because make
doesn't know if the conversion has taken place. Use a different name
for HAL before unmangling. Respect the build quietness settings.
Simplify Linux 2.4 rules so that the commands for decoding and
unmangling are written only once.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3815 0192ed92-7a03-0410-a25b-9323aeb14dbd
They appeared:
- in generic code and PARISC - in 2.6.12
- in PPC64 - in 2.6.14
- in merged PowerPC - in 2.6.15 (when the merge occurred)
- in MIPS - in 2.6.21
Improve comments and indentation, as the logic is getting more complex.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3712 0192ed92-7a03-0410-a25b-9323aeb14dbd
According to FreeBSD changelog, this area is present in 5416 and newer
and must be byteswapped on big-endian systems. Testing confirms it.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3701 0192ed92-7a03-0410-a25b-9323aeb14dbd
This should fix compilation on little-endian ARM and possibly other
platforms with little or no optimization. This closes#1783.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3700 0192ed92-7a03-0410-a25b-9323aeb14dbd
Remove kbuild by-products in regression/Makefile. Also allow
regression/tkip/test_tkip.c to compile (too few arguments to function
ieee80211_crypto_demic).
Closes: #1973
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3699 0192ed92-7a03-0410-a25b-9323aeb14dbd
Otherwise, $(TARGET).hal.o is not built for Linux 2.4 kernels. The rule
for ath_hal.o is for Linux 2.4 only and should not affect Linux 2.6
builds.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3604 0192ed92-7a03-0410-a25b-9323aeb14dbd
During the second stage of the module build, Linux 2.6.26 does not
define $(obj). Try using $(SUBDIRS) as $(obj) before falling back to
the current directory.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3603 0192ed92-7a03-0410-a25b-9323aeb14dbd
Filter out harmful options from OBJCOPY. Linux 2.4 adds "-O binary" and
"-S" for i386 and x86_64, which would break the build.
Use $(TOP) in hal_unmangle.objcopy path.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3599 0192ed92-7a03-0410-a25b-9323aeb14dbd
Normally, compatibility issues should be addressed in compat.h, but
INITIAL_JIFFIES is not something that should be widely used.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3598 0192ed92-7a03-0410-a25b-9323aeb14dbd
1) Move all debug preprocessor flags out into make files
2) Add support for remaining preprocessor flags to be set by ENV variables
3) Add support for HAL tracing to include device name
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3481 0192ed92-7a03-0410-a25b-9323aeb14dbd
both i386 and x86_64. When calculating TARGET for x86 platform, use
CONFIG_X86 and CONFIG_X86_64 rather than ARCH.
git-svn-id: http://madwifi-project.org/svn/madwifi/trunk@3318 0192ed92-7a03-0410-a25b-9323aeb14dbd
