Binutils ld has an annoying misfeature (apparently a regression from a
few years ago) that alignment directives (and alignment specifiers on
symbols) apply only to the runtime addresses and not, apparently, to
the load address region specified with the "AT>" syntax. The net
result is that by default the LMA output ends up too small for the
addresses generated in RAM. See here for some details:
https://sourceware.org/ml/binutils/2013-06/msg00246.htmlhttps://sourceware.org/ml/binutils/2014-01/msg00350.html
The required workaround/fix is that AFAICT any section which can have
inherit a separate VMA vs. LMA from a previous section must specify an
"ALIGN_WITH_INPUT" attribute. Otherwise the sections will get out of
sync and the XIP data will be wrong at runtime.
No, I don't know why this isn't the default behavior.
A further complexity is that this feature only works as advertised
when the section is declared with the "AT> region" syntax after the
block and not "AT(address)" in the header. If you use the header
syntax (with or without ALIGN_WITH_INPUT), ld appears to DOUBLE-apply
padding and the LMA ends up to big. This is almost certainly a
binutils bug, but it's trivial to work around (and the working syntax
is actually cleaner) so we adjust the usage here.
Note finally that this patch includes an effective reversion of commit
d82e9dd9 ("x86: HACK force alignment for _k_task_list section"), which
was an earlier workaround for what seems to be the same issue.
Jira: ZEP-955
Change-Id: I2accd92901cb61fb546658b87d6752c1cd14de3a
Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
Basic build framework for Nios2. Everything is stubbed out,
we just want to have a build going so that we can start to
parallelize implementation tasks.
This patch is not intended to be functional, but should be
able to produce a binary for all the nanokernel-based
sanity checks.
Change-Id: I12dd8ca4a2273f7662bee46175822c9bbd99202a
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
This no longer defaults to X86 and instead throws an error if a
supported arch isn't found. Output format now checks CONFIG_ARM
instead of Cortex M3, should be the same for all ARM chips.
Change-Id: Ib9bf07ad93a44b3f7cdac7b52432110776eb23fd
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Removed old style file description and documnetation and apply
doxygen synatx.
Change-Id: I3ac9f06d4f574bf3c79c6f6044cec3a7e2f6e4c8
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Change-Id: I6da43e41f9c6efee577b70513ec368ae3cce0144
Signed-off-by: Dan Kalowsky <daniel.kalowsky@intel.com>
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
These symbols are no longer required.
Change-Id: I99f6afad0ffb116efde1e2ff28252a2e4d8b7fb4
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Change all the Intel and Wind River code license from BSD-3 to Apache 2.
Change-Id: Id8be2c1c161a06ea8a0b9f38e17660e11dbb384b
Signed-off-by: Javier B Perez Hernandez <javier.b.perez.hernandez@linux.intel.com>
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: Benjamin Walsh <benjamin.walsh@windriver.com>
See
https://groups.google.com/forum/#!topic/ia32-abi/cn7TM6J_TIg
for more details.
• Support IA32 without FPU.
• Minimum ISA: Pentium ISA without x87 FPU instructions.
• Don't allow mixing i386 object files with Intel MCU object files.
• Support floating point with software emulation:
a. Long double is the same as double.
b. Use __float80 for 80-bit double.
• Minimize memory footprint:
a. Code size
b. Data size
c. Stack size
Here is the draft of Intel MCU psABI. The differences from IA32
psABI are
1. The minimum instruction set is Intel Pentium ISA minus instructions
for x87 floating point unit.
2. There are no x87 floating point registers.
3. There are no vector registers.
4. Segment registers are optional.
5. Support for TLS relocations are optional.
6. Scalar types larger than 4 bytes are aligned to 4 bytes.
7. There are no vector types.
8. _Decimal32, _Decimal64, and _Decimal128 types are optional.
9. long double type is the same as double.
10. float, double and long double types are passed and returned in
general purpose registers.
11. _Decimal32 and _Decimal64 types are passed in general purpose
registers.
12. Aggregate types no larger than 8 bytes are passed and returned
in general purpose registers.
13. Stack is 4-byte aligned.
14. The auxiliary vector support is optional.
15. Register %edx has undefined value at process entry.
16. New ELF machine code: EM_IAMCU.
17. New predefined C/C++ pre-processor symbols: __iamcu and __iamcu__
Change-Id: I6a0c45ad22d8f710b6f37a041aaa2fc1bf0b1c39
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
The change replaces multiple asterisks to ** at
the beginning of comments and adds a space before
the asterisks at the beginning of lines.
Change-Id: I7656bde3bf4d9a31e38941e43b580520432dabc1
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Makes name more consistent with other CPU_CORTEX_M* options.
Change-Id: I65968cb300207ba0de6231d9a67f2720be77b6ba
Signed-off-by: Peter Mitsis <peter.mitsis@windriver.com>
Not only is __start is a neutral label, but using a common entry point for all
architectures allows the entry point to be hoisted out of the linker script and
specified by the build system.
Change-Id: Ibcbb11a40ff964e0fc3c90afcfe231746071b040
Signed-off-by: Peter Mitsis <peter.mitsis@windriver.com>
Andy Ross