RISC-V newlib fails to build for soft float multilibs since long double
support was enabled in:
commit 04798b7bb6
Author: Kito Cheng <kito.cheng@sifive.com>
Date: Mon Dec 4 15:41:39 2023 +0800
RISC-V: Support long double math
Long double for RISC-V is using 128 bit IEEE 754 format like Aarch64,
so we reference AArch64 to support that.
The RISC-V soft floating point environment only supports the
FE_TONEAREST rounding mode and does not support exceptions. Guard long
double rounding and exception support with ifdefs based on the presence
of the relevant rounding modes and exceptions.
Tested on gcc/g++ testsuite using RISC-V GNU Newlib Toolchain built by
riscv-gnu-toolchain with multilibs:
riscv-sim/-march=rv32i/-mabi=ilp32/-mcmodel=medlow
riscv-sim/-march=rv32iac/-mabi=ilp32/-mcmodel=medlow
riscv-sim/-march=rv32im/-mabi=ilp32/-mcmodel=medlow
riscv-sim/-march=rv32imac/-mabi=ilp32/-mcmodel=medlow
riscv-sim/-march=rv32imafc/-mabi=ilp32f/-mcmodel=medlow
riscv-sim/-march=rv64imac/-mabi=lp64/-mcmodel=medlow
Co-authored-by: Simon Cook <simon.cook@embecosm.com>
These fixes fall into a few different buckets. First c99 doesn't allow a
parameter without a type. So in cases where the type had previously been an
implicit int, make it an explicit int. Second, for return values, don't allow
them to be implicit ints either, make them explicit. In a few cases change c89
function definitions to c99 function definitions. Lastly include <stdlib.h>
in sbrk.c to get the prototype for abort () which we call when we detect a
heap/stack collision.
By default, only split node html manuals are generated. Change the
default to single page, and add an extra rule to also generate the
split page manual.
Since $(AS) is the assembler, passing it a list of preprocessor include
flags doesn't make much sense. The files aren't preprocessed which means
`#include` lines aren't respected, and while it would affect `.include`
usage, we never use that, and it's extremely unlikely to change. Plus,
it's extremely unlikely we'd have .s files in common places to include vs
contained entirely within a specific arch dir, and at that point, it can
be included directly (with no flags), or the arch can add the unique set
of include paths that it needs for itself.
This reverts commit 17ac400c11.
The build failures were due to incorrectly using $(INCLUDES) when
running $(AS). Let's roll this back and drop $(INCLUDES) from the
$(AS) invocations.
gcc-14 will complain loudly both for calling a function without an in-scope
prototype or when the in scope prototype differs from the known signature.
"main" happens to be one of the functions the compiler knows about. So not
only do we need to prototype it, we need to make sure the prototype matches
what GCC thinks it should be.
This fixes the bfin libgloss port to do the right thing for bfin-elf.
Two libgloss issues specific to moxie-elf that show up build with GCC trunk.
First, putnum.c uses "print" without a prototype. So I added one based on the
function's definition in print.c. Second sim-inbyte.c uses read, which comes
from unistd.h, so this adds a suitable #include.
gcc-14 will default to c99 and as a result a fair amount of old code in newlib
(particularly libgloss) is failing to build. I don't offhand know how many
patches will be necessary to fix the various failures. I'll just pick them off
one by one from my tree.
This particular patch works around the return-mismatch problem syscalls.c for
fr30.
That file is a bit odd in that most functions are declared as returning an
integer, but the implementations look like:
> int
> _read (file, ptr, len)
> int file;
> char * ptr;
> int len;
> {
> asm ("ldi:8 %0, r0" :: "i" (SYS_read) : "r0");
> asm ("int #10");
>
> return;
> }
Note the lack of a value on the "return" statement. The assumption is that the
interrupt handler implementing syscalls will put the return value into the
proper register, so falling off the end of the C function or returning with no
value works in the expected way. It's not good code, but it probably works.
Working from that assumption I decided to just use a pragma to disable the
upgraded diagnostic from GCC -- essentially preserving existing behavior.
This is the only fr30 specific issue that needs to be resolved and the only
issue (so far) I've seen of this specific nature.
For the newlib part, warnings are all from lack of sync between
libc/machine/cris/sys/signal.h and libc/include/sys/signal.h. This
commit gets them sufficiently in sync again, functionality-wise and
declaration-wise. Still, nothing is declared that isn't supported at
the system level (i.e. in libgloss system calls and handled by the
CRIS simulator in the gdb project).
This is the libgloss part.
Recently, there was a change in gcc such that implicit function
declarations and type mismatches are now errors, no longer just
warnings. Fix by adding and correcting declarations warned about;
including the right header or adding a declaration (for "main" in
lcrt0.c) and adjust to those types as necessary.
- len must not be <= 0
- offset + len must not exceed off_t (max. file size)
Fixes: e01c50c7b0 ("Cygwin: introduce fallocate(2)")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
- len must not be <= 0
- offset + len must not exceed off_t (max. file size)
Fixes: 7636b58590 ("* autoload.cc (NtSetInformationFile): Define.")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Upstream GCC changed -Wimplicit-function-declaration warning into an
error. The build break about missing fpclassifyf function prototype
exposed a bug in the PRU port of libm.
The fix is to use the fpclassify macro for both double and float types.
Signed-off-by: Dimitar Dimitrov <dimitar@dinux.eu>
Upstream GCC has change this warning into error by default, so...we need
to explicitly convert the type from pointer from/to integer, generally
it's unsafe, but we know what we are doing here.
However it's not safe for ilp32 on RV64, but we didn't support that yet, so I
think this fix is good enough now :)
The out pointer is only used if data_chunk_count is > 0,
so there's no reason to set it to NULL in the error case.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
In cases where malloc fails for the 'g->matchjump' allocation, the code
path does not handle the failure gracefully, potentially leading to a
memory leak. This fix ensures proper cleanup by freeing the allocated
memory for 'pmatches' before returning.
Signed-off-by: Kuan-Wei Chiu <visitorckw@gmail.com>
Given that SSDs don't have a seek penalty, we can enable
automatic sparsifying of files on SSDs, even if the "sparse"
mount option is not set.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
In preparation of using the FileFsSectorSizeInformation info class,
add a couple of missing definitions.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Commit a3ae2a7348 ("Cygwin: don't autoload some kernel32 functions")
erroneously removed DiscardVirtualMemory from the list of autloaded
functions. DiscardVirtualMemory is not available on Windows 8.1.
Fixes: a3ae2a7348 ("Cygwin: don't autoload some kernel32 functions")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Replace multiplication with division for microseconds calculation from
nanoseconds in _gettimeofday function.
Signed-off-by: Kuan-Wei Chiu <visitorckw@gmail.com>
We've been manually editing newlib.hin after generating it with
autoheader to drop the version defines that we keep in the separate
_newlib_version.h header. This is confusing for people, and is an
easy source of mistakes/errors.
Since we're already running sed on newlib.h during configure to
filter out defines we don't want to expose, add the version macros
there too. This way we don't have to manually edit newlib.hin.
This simplifies the autoreconf step in exchange for a slightly more
complicated configure+sed step, but seems worth the trade-off.
Split fhandler_disk_file::fallocate into multiple methods, each
implementing a different aspect of fallocate(2), thus adding
FALLOC_FL_PUNCH_HOLE and FALLOC_FL_ZERO_RANGE handling.
For more correctly implementing posix_fallocate(3) semantics, make
sure to re-allocate holes in the given range if the file is sparse.
While at it, change the way checking when to make a file sparse.
The rule is now, make file sparse if the hole created by the action
spans at least one sparse block, taking the allocation granularity
of sparse files into account.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
write(2) sparsifies a file after an lseek far enough beyond EOF.
Let pwrite(2) sparsify as well if offset is far enough beyond EOF.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
fallocate is not supposed to return an errno code, it has to
return -1 and set errno.
Fixes: dd90ede40510 ("Cygwin: introduce fallocate(2)")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
open(2) implements O_TRUNC by just reducing the size of the file
to 0, to make sure EAs stay available.
Turns out, file sparseness is not removed this way either, so add
code to do just that.
Fixes: 603ef545bd ("* fhandler.cc (fhandler_base::open): Never open files with FILE_OVERWITE/FILE_OVERWRITE_IF.")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
First cut of the new, Linux-specific fallocate(2) function.
Do not add any functionality yet, except of basic handling
of FALLOC_FL_KEEP_SIZE.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
also, take mode flags parameter instead of just a bool.
Introduce __FALLOC_FL_TRUNCATE mode flag as internal flag to
indictae being called from ftruncate(2).
This is in preparation of an upcoming change introducing the
Linx-specific fallocate(2) call.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
The fhandler method ftruncate returns either EISDIR if it has been
called on directories, or EINVAL if called on files other than
regular files. This matches what ftruncate(2) is supposed to return,
but it doesn't match posix_fallocate(3), which is supposed to return
ENODEV in both cases.
To accomplish that, return ENODEV from fhandler_base::ftruncate()
and convert it to EINVAL in ftruncate(2). In posix_fallocate(3),
convert EISDIR to ENODEV.
Fixes: 7636b58590 ("* autoload.cc (NtSetInformationFile): Define.")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
The code introducing the lseek(2) code for the GNU extensions
SEEK_DATA and SEEK_HOLE accidentally checks if the mount point
has the "sparse" flag set and, if not, emulates SEEK_DATA/SEEK_HOLE
per the Linux specs.
However, the mount point "sparse" flag only determines whether
files should be made sparse or not. Files may be sparse independently
of that, obviously.
Fix that by checking for the FILE_ATTRIBUTE_SPARSE_FILE attribute
instead.
Fixes: edfa581d3c ("Cygwin: lseek: implement SEEK_DATA and SEEK_HOLE for files")
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Previously, e.g. nl_langinfo(_NL_TIME_WMONTH_1) returns "February"
due to the bug. Similarly, nl_langinfo(_NL_TIME_WWDAY_1) returns
"Mon". This occurs because wide char month and weekday arrays are
pointed off-by-one (e.g. the array wmon[12] is reffered as wmon[1-12]
rather than wmon[0-11]). This patch fixes that.
Fixes: d47d5b850b ("Extend locale support to maintain wide char values of native strings")
Reviewed-by: Corinna Vinschen <corinna@vinschen.de>
Signed-off-by: Takashi Yano <takashi.yano@nifty.ne.jp>
No longer drop ranges of identical link names. Append '#0, #1, ...'
to each name instead. Enhance charset allowed in label names.
No longer ignore null volume serial numbers.
Signed-off-by: Christian Franke <christian.franke@t-online.de>
Now that the low-level functions set and test stream orientation,
a few calls in API functions are redundant.
Signed-off-by: Corinna Vinschen <corinna@vinschen.de>
Jeff Law
Jeff Johnston
Craig Blackmore
Xiao Zeng
Mike Frysinger
Freddie Chopin
Hans-Peter Nilsson
Corinna Vinschen
Dimitar Dimitrov
Kito Cheng
Jon Turney
Kuan-Wei Chiu
Takashi Yano
Christian Franke
Sebastian Huber