- 23 Apr, 2014 40 commits
-
-
Anton Blanchard authored
Add some new definitions required to build the copyloop tests. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
Build the little endian ppc64 kernel with ABIv2 if the toolchain supports it. We can identify an ABIv2 capable toolchain by the -mabi=elfv2 compiler flag. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
__ftrace_make_call assumed ABIv1 TOC stack offsets, so it broke on ABIv2. While we are here, we can simplify the instruction modification code. Since we always update one instruction there is no need to probe_kernel_write and flush_icache_range, just use patch_branch. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
Now we have is_module_trampoline() and module_trampoline_target() we can remove a bunch of intimate kernel module trampoline knowledge from ftrace. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
ftrace has way too much knowledge of our kernel module trampoline layout hidden inside it. Create module_trampoline_target() that gives the target address of a kernel module trampoline. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
ftrace has way too much knowledge of our kernel module trampoline layout hidden inside it. Create is_module_trampoline() that can abstract this away inside the module loader code. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
Use ppc_function_entry in places where we previously assumed function descriptors exist. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
When testing the ftrace function tracer, I realised that ftrace_caller and mcount are called from modules and they both call into C, therefore they need the ABIv2 global entry point to establish r2. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
If an assembly function that calls back into c code is exported to modules, we need to ensure r2 is setup correctly. There are only two places crazy enough to do it (two of which are my fault). Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Rusty Russell authored
ELFv2 doesn't use function descriptors, because it doesn't need to load a new r2 when calling into a function. On the other hand, you're supposed to use a local entry point for R_PPC_REL24 branches. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Rusty Russell authored
ELFv2 doesn't need to set up r2 when calling a function. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Rusty Russell authored
In ELFv2, r12 is supposed to equal to PC on entry to a function. Our stubs use r11, so change swap that with r12. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Rusty Russell authored
ELFv2 uses a different stack offset (24 vs 40) to save r2. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Rusty Russell authored
ELFv2 doesn't use function descriptors, so we don't expect symbols to start with ".". But because depmod and modpost strip ".", and we have the special symbol ".TOC.", we still need to do it. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Rusty Russell authored
The new ELF ABI tends to use R_PPC64_REL16_LO and R_PPC64_REL16_HA relocations (PC-relative), so implement them. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Rusty Russell authored
The kernel resolved the '.TOC.' to a fake symbol, so we need to fix it up to point to our .toc section plus 0x8000. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Rusty Russell authored
For the ELFv2 ABI, powerpc introduces a magic symbol ".TOC.". If we don't create a CRC for it (minus the leading ".", since we strip that) we get a modpost warning about missing CRC and the CRC array seems to be displaced by 1 so other CRCs mismatch too. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Rusty Russell authored
For the ELFv2 ABI, powerpc introduces a magic symbol ".TOC.". depmod then complains that this doesn't resolve (so does modpost, but we could easily fix that). To export this, we need to use asm. modpost and depmod both strip "." from symbols for the old PPC64 ELFv1 ABI, so we actually export a "TOC.". Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Rusty Russell authored
Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Rusty Russell authored
By representing them as words, rather than chars, we can avoid endian ifdefs. Signed-off-by:
Rusty Russell <rusty@rustcorp.com.au>
-
Anton Blanchard authored
Don't try and dereference a function descriptor on ABIv2. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
There is no need to put a function descriptor in __secondary_hold_spinloop. Use ppc_function_entry to get the instruction address and put it in __secondary_hold_spinloop instead. Also fix an issue where we assumed cur_cpu_spec held a function descriptor. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
TRACE_WITH_FRAME_BUFFER creates 32 byte stack frames. On ppc64 ABIv1 this is too small and a callee could corrupt the stack by writing to the parameter save area (starting at offset 48). Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
The r2 TOC/GOT save offset is 40 on ABIv1 and 24 on ABIv2. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
Get rid of the tm specific STACK_PARAM and use STK_PARAM Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Ulrich Weigand authored
Some of the assembler files in lib/ make use of the fact that in the ELFv1 ABI, the caller guarantees to provide stack space to save the parameter registers r3 ... r10. This guarantee is no longer present in ELFv2 for functions that have no variable argument list and no more than 8 arguments. Change the affected routines to temporarily store registers in the red zone and/or the top of their own stack frame (in the space provided to save r31 .. r29, which is actually not used in these routines). In opal_query_takeover, simply always allocate a stack frame; the routine is not performance critical. Signed-off-by:
Ulrich Weigand <ulrich.weigand@de.ibm.com> Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
Fix STK_PARAM and use it instead of hardcoding ABIv1 offsets. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
Change how we setup registers for ret_from_kernel_thread. In ABIv1, instead of passing a function descriptor in, dereference it and pass the target in directly. Use ppc_global_function_entry to get it right on both ABIv1 and ABIv2. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
The MMU hashtable and SLB branch patching code uses function pointers for the update sites. This creates a difference between ABIv1 and ABIv2 because we don't have function descriptors on ABIv2. Get rid of the function pointer and just point at the update sites directly. This works on both ABIs. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
Replace FUNCTION_TEXT with ppc_function_entry which can handle both ABIv1 and ABIv2. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
Skip over the well known global entry point code for ABIv2. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
The linker fixes up .TOC. relocations, so prom_init_check.sh should ignore them. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
To establish addressability quickly, ABIv2 requires the target address of the function being called to be in r12. Fix a number of places in assembly code that we do indirect function calls. We need to avoid function descriptors on ABIv2 too. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
ABIv2 doesn't have function descriptors or dot symbols. One new thing it does add is a function global and a local entry point, so add that to our _GLOBAL macro. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
There are a few places we have to use dot symbols with the current ABI - the syscall table and the kvm hcall table. Wrap both of these with a new macro called DOTSYM so it will be easy to transition away from dot symbols in a future ABI. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
STD_EXCEPTION_COMMON, STD_EXCEPTION_COMMON_ASYNC and MASKABLE_EXCEPTION branch to the handler, so we can remove the explicit dot symbol and binutils will do the right thing. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
Now there are no users of _INIT_GLOBAL(), _STATIC() and _INIT_STATIC() we can remove them. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
There is no need to create a function descriptor for functions called locally out of assembly. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
There is no need to create a function descriptor for the system call table. By using one we force the system call table into the text section and it really belongs in the rodata section. This also removes another use of dot symbols. Signed-off-by:
Anton Blanchard <anton@samba.org>
-
Anton Blanchard authored
We have a number of places where we load the text address of a local function and indirectly branch to it in assembly. Since it is an indirect branch binutils will not know to use the function text address, so that trick wont work. There is no need for these functions to have a function descriptor so we can replace it with a label and remove the dot symbol. Signed-off-by:
Anton Blanchard <anton@samba.org>
-