Move the *INDENT-ON* annotation to the end of the file so that
uncrustify does not restyle the later sections (since it introduces a
risk of future problems).
Signed-off-by: David Horstmann <david.horstmann@arm.com>
The inline assembly defined in bn_mul.h confuses code style parsing,
causing code style correction to fail. Disable code style correction for
the whole section gated by "#if defined(MBEDTLS_HAVE_ASM)" to prevent
this.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
The following code:
#ifndef asm
#define asm __asm
#endif
causes Uncrustify to stop correcting the rest of the file. This may be
due to parsing the "asm" keyword in the definition.
Work around this by wrapping the idiom in an *INDENT-OFF* comment
wherever it appears.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
Fixes#1910
With ebx added to the MULADDC_STOP clobber list to fix#1550, the inline
assembly fails to build with GCC < 5 in PIC mode with the following error:
include/mbedtls/bn_mul.h:46:13: error: PIC register clobbered by ‘ebx’ in ‘asm’
This is because older GCC versions treated the x86 ebx register (which is
used for the GOT) as a fixed reserved register when building as PIC.
This is fixed by an improved register allocator in GCC 5+. From the release
notes:
Register allocation improvements: Reuse of the PIC hard register, instead of
using a fixed register, was implemented on x86/x86-64 targets. This
improves generated PIC code performance as more hard registers can be used.
https://www.gnu.org/software/gcc/gcc-5/changes.html
As a workaround, detect this situation and disable the inline assembly,
similar to the MULADDC_CANNOT_USE_R7 logic.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Within the M-profile of the Arm architecture, some instructions
admit both a 16-bit and a 32-bit encoding. For those instructions,
some assemblers support the use of the .n (narrow) and .w (wide)
suffixes to force a choice of instruction encoding width.
Forcing the size of encodings may be useful to ensure alignment
of code, which can have a significant performance impact on some
microarchitectures.
It is for this reason that a previous commit introduced explicit
.w suffixes into what was believed to be M-profile only assembly
in library/bn_mul.h.
This change, however, introduced two issues:
- First, the assembly block in question is used also for Armv7-A
systems, on which the .n/.w distinction is not meaningful
(all instructions are 32-bit).
- Second, compiler support for .n/.w suffixes appears patchy,
leading to compilation failures even when building for M-profile
targets.
This commit removes the .w annotations in order to restore working
code, deferring controlled re-introduction for the sake of performance.
Fixes#6089.
Signed-off-by: Hanno Becker <hanno.becker@arm.com>
Fixed a bug that the little-endian Microblaze does not work when MBEDTLS_HAVE_ASM is defined.
Signed-off-by: Kazuyuki Kimura <kim@wing.ocn.ne.jp>
Signed-off-by: Dave Rodgman <dave.rodgman@arm.com>
The variable is a local variable for the i386 bignum assembly only;
introduce it as part of the start/finish macros.
It can be noted that the variable is initialize to 0 within MULADDC_INIT,
so there are no data dependencies across blocks of MULADDC_INIT/CORE/STOP.
Signed-off-by: Hanno Becker <hanno.becker@arm.com>
Add memory constraints to the aarch64 inline assembly in MULADDC_STOP.
This fixes an issue where Clang 12 and 13 were generating
non-functional code on aarch64 platforms. See #4962, #4943
for further details.
Signed-off-by: David Horstmann <david.horstmann@arm.com>
MULADDC_CORE reads from (%%rsi) and writes to (%%rdi). This fragment is
repeated up to 16 times, and %%rsi and %%rdi are s and d on entry
respectively. Hence the complete asm statement reads 16 64-bit words
from memory starting at s, and writes 16 64-bit words starting at d.
Without any declaration of modified memory, Clang 12 and Clang 13 generated
non-working code for mbedtls_mpi_mod_exp. The constraints make the unit
tests pass with Clang 12.
Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>
Also remove preprocessor logic for MBEDTLS_CONFIG_FILE, since
build_info.h alreadyy handles it.
This commit was generated using the following script:
# ========================
#!/bin/sh
git ls-files | grep -v '^include/mbedtls/build_info\.h$' | xargs sed -b -E -i '
/^#if !?defined\(MBEDTLS_CONFIG_FILE\)/i#include "mbedtls/build_info.h"
//,/^#endif/d
'
# ========================
Signed-off-by: Bence Szépkúti <bence.szepkuti@arm.com>
These macros were moved into a header and now check-names.sh is failing.
Add an MBEDTL_ prefix to the macro names to make it pass.
Signed-off-by: Janos Follath <janos.follath@arm.com>
Move `include/mbedtls/bn_mul.h` to `library/bn_mul.h`.
Update includes and references to `bn_mul.h` to new location.
Also remove internal headers from `cpp_dummy_build.cpp` as it should only
test public headers in the library.
Signed-off-by: Chris Jones <christopher.jones@arm.com>
