In the initial code I missed one level of pointer indirection. Instead
of storing errno in impure_data, _impure_ptr was corrupted.
Only simulator is impacted. Real targets have no OS and no syscalls.
This resolves a bunch of stdio cases from the GCC testsuite:
FAIL->PASS: gcc.c-torture/execute/printf-2.c -O0 execution test
Signed-off-by: Dimitar Dimitrov <dimitar@dinux.eu>
Most code in newlib already uses unified syntax, but just a couple of
laggards remain. This patch removes these and means the the entire
code base has now been converted.
Previously, __internal_syscall() compiled into asm-code that unconditionally
sets the syscall argument registers a0 to a5.
For example, the instruction sequence for a exit syscall looked like
this:
li a0, 1 # in ther caller of exit()
# ... # in newlib:
li a1, 0 # unused arguments
li a2, 0
li a3, 0
li a4, 0
li a5, 0
li a7, 93 # exit syscall number
(i.e. the binary contains then 5 superfluous instructions for this
one argument syscall)
This commit changes the RISC-V syscall code such that only the required
syscall argument registers are set.
GCC detects that argc is known at compile time and thus evaluates all the
if-statements where argc is used at compile time (tested with -O2 and -Os).
When off_t is 32 bits, the value needs to be sign-extended to 64 bits
before shifting right to extract the high-order word. Previously
negative offsets were incorrectly encoded.
Signed-off-by: Sandra Loosemore <sandra@codesourcery.com>
The libm gamma functions use the _gamma_signgam field of the reentrant
structure, which changes offset with the --enable-newlib-reent-small
configure option, which means we need to use a newlib nano specific
version of libm in addition to libc in the nano.specs file. Reported
by Keith Packard. There is a riscv-gnu-toolchain patch that goes
along with this to create the new libm_nano.a file.
Signed-off-by: Jim Wilson <jimw@sifive.com>
This edits licenses held by Berkeley and NetBSD, both of which
have removed the advertising requirement from their licenses.
Signed-off-by: Keith Packard <keithp@keithp.com>
Target libraries are considered to be built for GCC's "host", not GCC's
"target". The "host" variable must be set by configure scripts using
"config-ml.in" to determine multilib support, otherwise disabled
multilibs (specified as a configure argument with --disable-<multilib>)
will still be built for the subdirectories those configure scripts
reside in.
Commit 72ff9acad2 caused st_atime,
st_ctime, and st_mtime to be defined as macros. This collided with
use of these identifiers as field names in struct gdb_stat (which
represents the GDB RSP encoding of struct stat) in libgloss
semihosting support for nios2 and m68k. This patch renames the
affected fields of struct gdb_stat.
Signed-off-by: Sandra Loosemore <sandra@codesourcery.com>
- revert previous fix which altered sys/stat.h
- fix libgloss/cris/gensyscalls to undef st_atime, st_mtime,
and st_ctime macros which cannot be used with new_stat structure
Applied changes from commit 8d98f95:
* arm/crt0.S: Initialise __heap_limit when ARM_RDI_MONITOR is defined.
* arm/syscalls.c: define __heap_limit global symbol.
* arm/syscalls.c (_sbrk): Honour __heap_limit.
Applied changes from commit 8d98f95:
Fixed semihosting for ARM when heapinfo not provided by debugger
Left-over part of commit 84b2a020da
The _end marker must be aligned to 4-bytes to ensure that the last
element written does not reach beyond the address of _end. This is
also necessary as the termination condition is an equality test
instead of an ordered test so (_end - _fbss) must be a multiple of
4-bytes. The alignment is already correct for mti*.ld files, fix
it for all remaining MIPS scripts that don't already align to at
least 4.
The .init/.fini sections are not required for msp430-elf, and add unnecessary
code bloat to the CRT library. These sections are specified as "unused" by
the MSP430 EABI.
.init existed to call __crt0_run_{init,preinit}_array which run through
the functions in .{init,preinit}_array.
__crt0_run_{init,preinit}_array are already dynamically included like the
other crt0 functions, so these can be placed before the call to main,
which ensures they are still called if needed.
With these functions moved, .init has no purpose and can be removed.
.fini existed to call __crt0_run_fini_array.
However, the "__msp430_fini" symbol which marks the start of .fini has
never been used, so no termination routines have ever been run for
msp430. On returning from main(), _exit() is called which just loops
forever.
So there is no current expectation that __crt0_run_fini_array will
get called by the CRT code. Further work is to ensure functions
registered with atexit can be optionally called during program termination,
and then __crt0_run_fini_array can be registered with atexit during
program initialization.
The mechanisms for supporting the "-minrt" option have also been removed.
"-minrt" enabled a "minimum runtime environment" by removing calls to
functions which run global static initializers and constructors. Since
this behaviour is now dynamic, and these functions are only included
when needed, the minrt versions of the CRT object files are no longer
required.
SP initialization changes:
1. set default value in semihosting case as well
2. moved existing SP & SL init code for processor modes in separate routine and made it as "hook"
3. init SP for processor modes in Thumb mode as well
Add new macro FN_RETURN, FN_EH_START and FN_EH_END.
This patch adds _LITE_EXIT in crt0.S to enable "lite exit" technique in
RISC-V. The changes have been tested in riscv/riscv-gnu-toolchain by
riscv-dejagnu with riscv-sim.exp/riscv-sim-nano.exp.
The _fdata symbol in MIPS linker scripts is aligned to a 16-byte
boundary. The ALIGN function does not implicitly update current
location counter. If sections positioned after the assignment
do not have the same natural alignment as the ALIGN function then
the start of the section group will not coincide with the value
of the symbol.
Given the linker command sequence:
symbol = ALIGN (NN);
(.section*)
where the idiom implies a desire to mark the beginning of .section
with symbol, there must be an assignment to the location counter
between the assignment to symbol and the .section pattern.
libgloss/
* mips/array.ld: Update the location counter to match _fdata.
* mips/cfe.ld: Likewise.
* mips/ddb-kseg0.ld: Likewise.
* mips/ddb.ld: Likewise.
* mips/dve.ld: Likewise.
* mips/idt.ld: Likewise.
* mips/idt32.ld: Likewise.
* mips/idt64.ld: Likewise.
* mips/idtecoff.ld: Likewise.
* mips/jmr3904app-java.ld: Likewise.
* mips/jmr3904app.ld: Likewise.
* mips/jmr3904dram-java.ld: Likewise.
* mips/jmr3904dram.ld: Likewise.
* mips/lsi.ld: Likewise.
* mips/mti32.ld: Likewise.
* mips/mti64.ld: Likewise.
* mips/mti64_64.ld: Likewise.
* mips/mti64_n32.ld: Likewise.
* mips/nullmon.ld: Likewise.
* mips/pmon.ld: Likewise.
* mips/sde32.ld: Likewise.
* mips/sde64.ld: Likewise.
The compiler driver positions the linker script at the end of the linker
command-line, after crtend.o. As a result, any INPUT objects and archive
GROUPs introduced by the linker script are placed after crtend.o and the
end-of-frame marker provided by crtend.o ends up in between .eh_frames
instead of being at the end.
This has always been a problem, but a binutils update to clean-up
redundant NULL markers in .eh_frame exposes it as a execution failure in
exception-handling tests. This patch re-orders .eh_frames in all
MIPS linker scripts so that the one from crtend.o is always placed last.
libgloss/
* mips/array.ld: Re-order to place .eh_frame from crtend.o
after all other .eh_frame sections.
* mips/cfe.ld: Likewise.
* mips/ddb-kseg0.ld: Likewise.
* mips/ddb.ld: Likewise.
* mips/dve.ld: Likewise.
* mips/idt.ld: Likewise.
* mips/idt32.ld: Likewise.
* mips/idt64.ld: Likewise.
* mips/jmr3904app.ld: Likewise.
* mips/lsi.ld: Likewise.
* mips/mti32.ld: Likewise.
* mips/mti64.ld: Likewise.
* mips/mti64_64.ld: Likewise.
* mips/mti64_n32.ld: Likewise.
* mips/nullmon.ld: Likewise.
* mips/pmon.ld: Likewise.
* mips/sde32.ld: Likewise.
* mips/sde64.ld: Likewise.
Many of the MSP430 crt functions (e.g. to initialize bss) are linked
"dynamically", based on symbols defined in the program.
The GNU assembler defines the symbols corresponding to the crt
functions by examining the section names in the input file.
If GCC has been configured with --enable-initfini-array, then
.init_array and .fini_array will hold pointers to global
constructors/destructors. These sections can also hold functions that
need to be executed for other purposes.
The attached patch puts the __crt0_run_{preinit,init,fini}_array and
__crt0_run_array functions in their own object files, so they will
only be linked when needed.
Successfully regtested the DejaGNU GCC testsuite using the binutils and
newlib changes together with GCC trunk configured with
--enable-initfini-array.
The code in trap.S is to support the old APCS chunked stack variant,
which dates back to the Acorn days, so put it under #ifndef
__ARM_EABI__.
* libgloss/arm/trap.S: Use __ARM_EABI rather than PREFER_THUMB.
* newlib/libc/sys/arm/trap.S: Use __ARM_EABI rather than
__thumb2__.
The location of the handler at offset 0x20 from the start of memory,
immediately after the 32-byte reset vector, matches the expectations
of real hardware (e.g., a 3c120 board).
Author: Sandra Loosemore <sandra@codesourcery.com>
Date: Wed Mar 13 20:22:16 2019 -0700
Add semihosting documentation for nios2 and m68k.
QEMU maintainers have asked for a specification of the nios2
semihosting interface. Since it's essentially a copy of the m68k
implementation, this patch adds a document for that target as well.
The M class cores don't support Semihosting v2 mixed mode, but we were
accidentally using the new immediates for it. My last patch changed the
immediates which broke the build because doing a full multi-lib build
including M architectures now results in an assembler error instead of
silently doing the wrong thing.
This fixes the issue by changing the defines around such that According
to the specs any M class build uses the normal semihosting instructions.
Regtested on arm-none-eabi and no issues, using a build with m class
multilibs too.
The Semihosting v2 protocol requires us to output the Armv8-a HLT instruction
when in mixed mode (SEMIHOST_V2_MIXED_MODE), however it also requires this to
be done for Armv7-a and earlier architectures.
The HLT instruction is defined in the undefined encoding space for older
architectures but simulators such as QEMU already trap on it [1] for all
architectures and is a requirement for semihosting v2 [2].
Unfortunately the GAS restricts the use of HLT to Armv8-a which requires us to
use the instruction encodings we want directly in crt0.
This patch does this, I have not updated newlib/libc/* as that is quite out of
date already. A proper sync is needed in order to get things back in sync.
A different patch for this would be best.
[1] 19a6e31c9d
[2] https://developer.arm.com/docs/100863/latest/the-semihosting-interface
The toplevel makefile used by binutils/gcc/newlib/etc has install-pdf and
install-html targets, but they fail because libgloss doesn't support them.
Tested with an arm-eabi combined tree build and install, and verifying that
the install-pdf and install-html targets now work, and that the pdf and html
doc files are now in the install tree.
libgloss/
* Makefile.in (install-html, install-pdf): New.
* doc/Makefile.in (htmldir, pdfdir): New.
(porting.ps): Delete white space on blank line.
(install-pdf, install-html): New.
The _exit function currently passes -1 as a "sig" to the _kill function as an
invalid signal number so that _kill can distinguish between an abort and a
standard exit.
For boards using the SYS_EXIT_EXTENDED semi-hosting operation to return a
status code, this means that the "status" paramter to _exit is ignored and the
return code is always -1.
https://developer.arm.com/docs/100863/latest/semihosting-operations/sys_exit_extended-0x20
This patch puts shared code between _kill and _exit into a new function
_kill_shared that takes the semi-hosting "reason" to use (if semi-hosting is
available) as an argument.
For semi-hosting _kill_shared provides that "reason".
Without the "sig" argument being used to distinguish between a normal and
abnormal exit, the _exit function can provide the return code to be used if the
SYS_EXIT_EXTENDED operation is available.
Hence the exit code can be returned.
This patch is similar the arm one committed recently.
2018-10-08 Christophe Lyon <christophe.lyon@linaro.org>
* libgloss/aarch64/syscalls.c (_sbrk): Fix prototype.
(_getpid, _write, _swiwrite, _lseek, _swilseek, _read, _wriread):
Likewise.
AngelSWI_Reason_ReportException does not return accoring to the ARM
documentation, so it is valid to mark _kill() as noreturn. This way,
the compiler does not warn about _exit() returning a value despite
being noreturn.
2018-10-01 Christophe Lyon <christophe.lyon@linaro.org>
* libgloss/arm/_exit.c (_exit): Declare _kill() as noreturn.
* libgloss/arm/_exit.c (_kill): Likewise. Remove the return
statements.
* newlib/libc/sys/arm/syscalls.c (_kill): Likewise..
It's been a while... I see the CRIS port broke with the
time_t-default-to-64-bit change, observable by a few test-cases in the
gcc fortran(!) tests failing, regressing when trying a recent newlib.
This is a two-part belt-and-suspenders change: adjust the CRIS port
gettimeofday syscall (the only one in newlib/CRIS passing a time_t or
struct timeval) to handle a userspace 64-bit time_t and secondly default
time_t to 32-bit long anyway. I considered making the local
"kernel_timeval" copy in _gettimeofday conditional on (userspace) time_t
being 64 bits, but thought it not worth bothering with the few move insns.
The effect of a 64-bit time_t is however observable as longer simulation
time when running the gcc testsuite and as bigger binaries without any
actual upside from the larger time_t size, so I thought better make the
default for this port go back to being a "long" again.
Tested by running the gcc testsuite over the three combinations of two
parts of the patch and observing the expected changes. Committed.
libgloss:
Adjust for syscall and userspace having different time_t or timeval.
* cris/linunistd.h (kernel_time_t, kernel_suseconds_t, kernel_timeval):
New types.
(gettimeofday): Change the type of the first argument to be a
pointer to a struct kernel_timeval.
* cris/gensyscalls (_gettimeofday): Use an intermediate struct
kernel_timeval for the syscall and initialize the result from
that.
Signed-off-by: Hans-Peter Nilsson <hp@axis.com>
Upon successful completion, times() shall return the elapsed real time,
in clock ticks, since an arbitrary point in the past (for example,
system start-up time).
Signed-off-by: Kito Cheng <kito.cheng@gmail.com>
Introduce new host configuration variable "have_init_fini" which is set
to "yes" by default. Override it for RISC-V to "no".
Signed-off-by: Sebastian Huber <sebastian.huber@embedded-brains.de>
The current SYS_EXIT has a bug that when making the call it always uses
the v2 calling convention. This is undefined behavior according to the
semihosting specification:
https://developer.arm.com/docs/100863/latest/semihosting-operations/sys_exit-0x18
This patch fixes it by making sure v1 passes the argument directly in the register instead
of in a block. And for v2 it does the same if the v2 extension isn't supported.
The sequence generated now is
12424: ebfffecd bl 11f60 <_has_ext_exit_extended>
12428: e3500000 cmp r0, #0
1242c: 11a0500d movne r5, sp
12430: 059d5000 ldreq r5, [sp]
12434: e1a00004 mov r0, r4
12438: e1a01005 mov r1, r5
1243c: ef00f000 svc 0x0000f000
Signed-off-by: Tamar Christina <tamar.christina@arm.com>
Previously, hw exception handler stub and interrupt handler stub for microbaze were unable to
be overwritten. Change to weak to fix this.
Signed-off-by: Ben Levinsky <ben.levinsky@xilinx.com>
Dimitar Dimitrov
Richard Earnshaw
Georg Sauthoff
Sandra Loosemore
Jim Wilson
Keith Packard
Jozef Lawrynowicz
Anthony Green
Jeff Law
Jeff Johnston
Alexander Fedotov
Faraz Shahbazker
Faraz Shahbazker
Christophe Lyon
Tamar Christina
Matthew Malcomson
Hans-Peter Nilsson
Denis Ivanov
Sebastian Huber
Ben Levinsky
Jaap de Wolff
Chih-Mao Chen
Yaakov Selkowitz