Message ID | 1442569912-13452-1-git-send-email-judge.packham@gmail.com |
---|---|
State | Superseded |
Headers | show |
Dear Chris Packham, On 09/18/2015 10:51 AM, Chris Packham wrote: > From the micropython source: > > This mpz module implements arbitrary precision integers. > > The storage for each digit is defined by mpz_dig_t. The actual number > of bits in mpz_dig_t that are used is defined by MPZ_DIG_SIZE. The > machine must also provide a type that is twice as wide as mpz_dig_t, > in both signed and unsigned versions > > MPZ_DIG_SIZE can be between 4 and 8*sizeof(mpz_dig_t), but it makes > most sense to have it as large as possible. > > Micropython detects x86_64 targets and sets MPZ_DIG_SIZE appropriately > but for other 64-bit targets we need to explicitly pass > -DMPZ_DIG_SIZE=32 in CFLAGS. > > Signed-off-by: Chris Packham <judge.packham@gmail.com> > --- > package/micropython/micropython.mk | 4 ++++ > 1 file changed, 4 insertions(+) > > diff --git a/package/micropython/micropython.mk b/package/micropython/micropython.mk > index 6266882..1ac2e8d 100644 > --- a/package/micropython/micropython.mk > +++ b/package/micropython/micropython.mk > @@ -16,6 +16,10 @@ ifeq ($(BR2_powerpc)$(BR2_sh)$(BR2_xtensa),y) > MICROPYTHON_CFLAGS = -DMICROPY_GCREGS_SETJMP=1 > endif > > +ifeq ($(BR2_ARCH_IS_64),y) > +MICROPYTHON_CFLAGS += -DMPZ_DIG_SIZE=32 > +endif > + you have to take into account that MIPS64 n32 is a 64-bit target as well, but it has a MP_SSIZE_MAX of 32-bit. So, if you set MPZ_DIG_SIZE to 32 for all 64-bit arches, then you will see the same failure for the when building for MIPS64 n32. I suggest you to do modify the "ifeq" like this: # MIPS64 n32 is a 64-bit arch, but it has a 32-bit wide SSIZE_MAX which # means we still need the default MPZ_DIG_SIZE of 16. ifeq ($(BR2_ARCH_IS_64)$(BR2_MIPS_NABI32),y) Regards, Vincent. > define MICROPYTHON_BUILD_CMDS > $(TARGET_MAKE_ENV) $(MAKE) -C $(@D)/unix \ > CROSS_COMPILE=$(TARGET_CROSS) \ >
Hi Vincent, On Fri, Sep 18, 2015 at 10:02 PM, Vicente Olivert Riera <Vincent.Riera@imgtec.com> wrote: > Dear Chris Packham, > > On 09/18/2015 10:51 AM, Chris Packham wrote: >> From the micropython source: >> >> This mpz module implements arbitrary precision integers. >> >> The storage for each digit is defined by mpz_dig_t. The actual number >> of bits in mpz_dig_t that are used is defined by MPZ_DIG_SIZE. The >> machine must also provide a type that is twice as wide as mpz_dig_t, >> in both signed and unsigned versions >> >> MPZ_DIG_SIZE can be between 4 and 8*sizeof(mpz_dig_t), but it makes >> most sense to have it as large as possible. >> >> Micropython detects x86_64 targets and sets MPZ_DIG_SIZE appropriately >> but for other 64-bit targets we need to explicitly pass >> -DMPZ_DIG_SIZE=32 in CFLAGS. >> >> Signed-off-by: Chris Packham <judge.packham@gmail.com> >> --- >> package/micropython/micropython.mk | 4 ++++ >> 1 file changed, 4 insertions(+) >> >> diff --git a/package/micropython/micropython.mk b/package/micropython/micropython.mk >> index 6266882..1ac2e8d 100644 >> --- a/package/micropython/micropython.mk >> +++ b/package/micropython/micropython.mk >> @@ -16,6 +16,10 @@ ifeq ($(BR2_powerpc)$(BR2_sh)$(BR2_xtensa),y) >> MICROPYTHON_CFLAGS = -DMICROPY_GCREGS_SETJMP=1 >> endif >> >> +ifeq ($(BR2_ARCH_IS_64),y) >> +MICROPYTHON_CFLAGS += -DMPZ_DIG_SIZE=32 >> +endif >> + > > you have to take into account that MIPS64 n32 is a 64-bit target as > well, but it has a MP_SSIZE_MAX of 32-bit. So, if you set MPZ_DIG_SIZE > to 32 for all 64-bit arches, then you will see the same failure for the > when building for MIPS64 n32. > > I suggest you to do modify the "ifeq" like this: > > # MIPS64 n32 is a 64-bit arch, but it has a 32-bit wide SSIZE_MAX which > # means we still need the default MPZ_DIG_SIZE of 16. > ifeq ($(BR2_ARCH_IS_64)$(BR2_MIPS_NABI32),y) > I'll take a look. The remote machine I was doing my buildroot hacking on has gone AWOL so it may take a while for me to get it back up and running (there was a planned server outage so I suspect it's trying to access some non-existent mounts). When I asked upstream there was a thought that this commit[1] might fix the issue. So I'll try adding that as a patch first. -- [1] - https://github.com/micropython/micropython/commit/8b4fb4fe140e9cf57fcfa258d0d2d6fe19090fc5
diff --git a/package/micropython/micropython.mk b/package/micropython/micropython.mk index 6266882..1ac2e8d 100644 --- a/package/micropython/micropython.mk +++ b/package/micropython/micropython.mk @@ -16,6 +16,10 @@ ifeq ($(BR2_powerpc)$(BR2_sh)$(BR2_xtensa),y) MICROPYTHON_CFLAGS = -DMICROPY_GCREGS_SETJMP=1 endif +ifeq ($(BR2_ARCH_IS_64),y) +MICROPYTHON_CFLAGS += -DMPZ_DIG_SIZE=32 +endif + define MICROPYTHON_BUILD_CMDS $(TARGET_MAKE_ENV) $(MAKE) -C $(@D)/unix \ CROSS_COMPILE=$(TARGET_CROSS) \
From the micropython source: This mpz module implements arbitrary precision integers. The storage for each digit is defined by mpz_dig_t. The actual number of bits in mpz_dig_t that are used is defined by MPZ_DIG_SIZE. The machine must also provide a type that is twice as wide as mpz_dig_t, in both signed and unsigned versions MPZ_DIG_SIZE can be between 4 and 8*sizeof(mpz_dig_t), but it makes most sense to have it as large as possible. Micropython detects x86_64 targets and sets MPZ_DIG_SIZE appropriately but for other 64-bit targets we need to explicitly pass -DMPZ_DIG_SIZE=32 in CFLAGS. Signed-off-by: Chris Packham <judge.packham@gmail.com> --- package/micropython/micropython.mk | 4 ++++ 1 file changed, 4 insertions(+)