Jeffrey Walton <[email protected]> writes: > Compile results in the following. According to Stallman I believe you > are supposed to add the necessary options to compile the file > independent of CFLAGS, and it should not need to be override-able by > the user because it is required. Also see > https://www.gnu.org/prep/standards/html_node/Command-Variables.html .
Note that section talks about make variables, and overriding them like make all CFLAGS=... It's not talking about the configure settings with the same name. GMP uses automake, and I don't know if automake-generated Makefiles conform to that part of the GNU coding standads (it uses a bunch of different *_CFLAGS make variables for different purposes), or if gmp's handling of CFLAGS departs from automake conventions. Anyway, to GMP's configure script, setting CFLAGS simply means "Just do as I say; I know what I'm doing". So please don't specify CFLAGS that makes the compiler not support the intended target platform. > Assembler messages: > tmp-lshift.s:106: Error: selected processor does not support ARM mode > `vdup.32 d6,r3' You don't tell us your intention. If you want to include neon code in the build, and you specify CFLAGS yourself, you have to include whatever options are needed to make the assembler recognize them. Or if you don't want neon code, you must specify a configuration without neon support, to exclude use of assembly files depending on neon. For the specific case of neon, one alternative might be to explicitly add the pseudo-op .fpu neon in all neon asm files, which should make them less dependent on flags passed on the compiler command line. It looks like the definition of ASM_START in mpn/arm/arm-defs.m4 can add that for us. So should we be using ASM_START(`neon') in all assembly files targetting neon? Regards, /Niels -- Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677. Internet email is subject to wholesale government surveillance. _______________________________________________ gmp-bugs mailing list [email protected] https://gmplib.org/mailman/listinfo/gmp-bugs
