-Wno-address-of-packed-member -Wno-stringop-truncation and -Wno-array-bounds
are needed to be able to build with sid, but are not supported by etch
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Changing the type of offset to off_t has implictely pulled
the use of __divdi3()//__moddi3() that we are not importing from
the system. Cast offset to int to avoid this
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
To build inside an m68k/sid env:
./run-sid ./configure
./run-sid make all
m68k/etch env:
./run-etch ./configure
./run-etch make all
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
The switch statements in emile_scsi_get_dev() and emile-set-output
were missing break statements:
* MAJOR_IDE1 in emile_scsi_get_dev() fell through to default,
resulting in an "Unknown device major number" error.
* ARG_STOPBITS in emile-set-output fell through, so subsequent code
executed within the switch block.
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
When restoring a bootblock, the path .backup should be renamed to
.backup.old. The sprintf() call mistakenly used the newly allocated
buffer rather than the original path. We now pass backup_path to
sprintf() before renaming the backup file.
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
Including the stack pointer register in a clobber list has long since been deprecated by GCC (~9.0 and ~8.4), and according to my research, it appears the compiler ignores its presence in a clobber list regardless.
Whether this is the correct thing to do here I do not know, but from my research, it seems to be, and it does silence the associated deprecation warning that, due to -Werror, causes builds to fail.
See here: https://stackoverflow.com/questions/64928418/arm-cortex-m0-set-msp-c17-compiler-warning
This file has not existed in the repository since 2017, which means attempting to make dist will fail with an error for it being missing. This removes the reference to that file, allowing make dist to succeed.
Considering configure has to be run first (and is what tools.sh became back in 2017), I do not expect the complete omission of this reference to cause any issues.
The former macro has long since been removed from upstream, causing build failures on modern tools. This was fixed in ext4magic all the way back in 2018, for example.
See here: https://sourceforge.net/p/ext4magic/tickets/10/
by default ar uses deterministic mode and all .o are
added even if they are not new. Disable this by adding -U
to ARFLAGS.
Signed-off-by: Laurent Vivier <laurent@vivier.eu>
