Date: Tue, 23 Dec 2014 13:28:33 -0700 From: Ian Lepore <ian@freebsd.org> To: Larry Rosenman <ler@lerctr.org> Cc: freebsd-arm@freebsd.org Subject: Re: lsof on FreeBSD/arm Message-ID: <1419366513.1018.141.camel@freebsd.org> In-Reply-To: <c9bcab9ff5aa4a67ec290908e72a53c9@thebighonker.lerctr.org> References: <dfbaf5451abce25b512efe4f8c80ff58@thebighonker.lerctr.org> <1419362189.1018.138.camel@freebsd.org> <c9bcab9ff5aa4a67ec290908e72a53c9@thebighonker.lerctr.org>
index | next in thread | previous in thread | raw e-mail
[-- Attachment #1 --] On Tue, 2014-12-23 at 14:02 -0600, Larry Rosenman wrote: > On 2014-12-23 13:16, Ian Lepore wrote: > > On Tue, 2014-12-23 at 11:29 -0600, Larry Rosenman wrote: > >> Greetings, > >> I'm the maintainer of the sysutils/lsof port for FreeBSD. It's > >> been > >> reported to me that the port doesn't compile on arm. In investigating > >> that, I found that some of the includes that lsof includes need a cpu > >> type defined in order to compile (machine/cpuconf.h). > >> > >> Is there a reason that there is no "default" cpu defined either by the > >> compiler or environment? > >> > >> Playing in a armv6 poudriere jail, I found I can get it to compile > >> via: > >> borg.lerctr.org /usr/local/etc/poudriere.d $ cat p110armv6-make.conf > >> CFLAGS += -DCPU_ARM1136=1 > >> borg.lerctr.org /usr/local/etc/poudriere.d $ > >> > >> However, I'm not sure if that's kosher for a port to just define. > >> > >> Can any of the arm experts help here? > >> > >> > >> > > > > See if the attached patch helps. The specific type of cpu shouldn't > > really matter outside the kernel (or even inside it except in a couple > > places), what usually matters is the architecture, which we've > > simplified to v4 versus v6 for most purposes. > > > > The attached patch will use the compiler's predefined values for arch > > to > > set our old-school symbols, which we really need to do a big cleanup > > pass on, but hopefully this little hack will get you going until that > > gets done. Let me know if this works and I'll commit it. > > > > -- Ian > Thanks. I'm having a hard time convincing poudriere to apply this patch > to the environment > > Any ideas? Heh, well, it would help if I had tested the patch first... I'm having a scatterbrain day today. Try this one. In fact, I'll just attach the cpuconf.h itself too so that you can just copy it into the emulator filesystem for a quick test. -- Ian [-- Attachment #2 --] Index: sys/arm/include/cpuconf.h =================================================================== --- sys/arm/include/cpuconf.h (revision 274850) +++ sys/arm/include/cpuconf.h (working copy) @@ -99,6 +99,25 @@ #endif #define ARM_NARCH (ARM_ARCH_4 + ARM_ARCH_5 + ARM_ARCH_6 | ARM_ARCH_7A) + +/* + *Compatibility for userland builds that have no CPUTYPE defined. Use the + *symbols predefined by the compiler to get into the right arch ballpark. + */ +#if ARM_NARCH == 0 +#if defined(__ARM_ARCH_4T__) +#undef ARM_ARCH_4 +#undef ARM_NARCH +#define ARM_ARCH_4 1 +#define ARM_NARCH 1 +#elif defined(__ARM_ARCH_6ZK__) +#undef ARM_ARCH_6 +#undef ARM_NARCH +#define ARM_ARCH_6 1 +#define ARM_NARCH 1 +#endif +#endif + #if ARM_NARCH == 0 && !defined(KLD_MODULE) && defined(_KERNEL) #error ARM_NARCH is 0 #endifhelp
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1419366513.1018.141.camel>