Message ID | 20160312031535.GA24351@home.buserror.net (mailing list archive) |
---|---|
State | Accepted |
Headers | show |
On Fri, 2016-03-11 at 21:15 -0600, Scott Wood wrote: > Highlights include 8xx optimizations, 32-bit checksum optimizations, 86xx > consolidation, e5500/e6500 cpu hotplug, more fman and other dt bits, and > minor fixes/cleanup. Hi Scott, This one's giving me a few troubles. > chenhui zhao (6): > powerpc/mm: any thread in one core can be the first to setup TLB1 > powerpc/cache: add cache flush operation for various e500 > powerpc/rcpm: add RCPM driver That commit doesn't build with SMP=n, as reported by Guenter: arch/powerpc/sysdev/fsl_rcpm.c:32:2: error: implicit declaration of function 'get_hard_smp_processor_id' [-Werror=implicit-function-declaration] That's not a biggy, I can probably fix it up here. > powerpc/mpc85xx: Add CPU hotplug support for E6500 And this one breaks with old binutils (<= 2.22) because of: arch/powerpc/kernel/head_64.S:210: Error: Unrecognized opcode: `mttmr' If you're happy to break the build on older bintuils for those configs then that's OK with me. > Christophe Leroy (31): > powerpc/8xx: Handle CPU6 ERRATA directly in mtspr() macro This breaks mpc866_ads_defconfig for me, with lots of: arch/powerpc/mm/8xx_mmu.c:139:2: error: memory input 1 is not directly addressable That's using gcc < 4.9. With >= 4.9 it builds OK. So again your call on whether we drop support for those compilers for that config. cheers
On Tue, 2016-03-15 at 11:19 +1100, Michael Ellerman wrote: > On Fri, 2016-03-11 at 21:15 -0600, Scott Wood wrote: > > > Highlights include 8xx optimizations, 32-bit checksum optimizations, 86xx > > consolidation, e5500/e6500 cpu hotplug, more fman and other dt bits, and > > minor fixes/cleanup. > > Hi Scott, > > This one's giving me a few troubles. > > > chenhui zhao (6): > > powerpc/mm: any thread in one core can be the first to setup TLB1 > > powerpc/cache: add cache flush operation for various e500 > > powerpc/rcpm: add RCPM driver > > That commit doesn't build with SMP=n, as reported by Guenter: > > arch/powerpc/sysdev/fsl_rcpm.c:32:2: error: implicit declaration of > function 'get_hard_smp_processor_id' [-Werror=implicit-function-declaration] > > That's not a biggy, I can probably fix it up here. By adding asm/smp.h? I missed this since we don't have a non-SMP defconfig that includes this. > > powerpc/mpc85xx: Add CPU hotplug support for E6500 > > And this one breaks with old binutils (<= 2.22) because of: > > arch/powerpc/kernel/head_64.S:210: Error: Unrecognized opcode: `mttmr' > > If you're happy to break the build on older bintuils for those configs then > that's OK with me. I'm not too bothered, but it's also an easy fix -- patch sent. > > Christophe Leroy (31): > > powerpc/8xx: Handle CPU6 ERRATA directly in mtspr() macro > > This breaks mpc866_ads_defconfig for me, with lots of: > > arch/powerpc/mm/8xx_mmu.c:139:2: error: memory input 1 is not directly > addressable > > That's using gcc < 4.9. With >= 4.9 it builds OK. So again your call on > whether > we drop support for those compilers for that config. I'd rather not require something that new, especially for a chip so old. I've sent an untested patch that should fix it. -Scott
On Tue, 2016-03-15 at 02:01 -0500, Scott Wood wrote: > On Tue, 2016-03-15 at 11:19 +1100, Michael Ellerman wrote: > > On Fri, 2016-03-11 at 21:15 -0600, Scott Wood wrote: > > > > > Highlights include 8xx optimizations, 32-bit checksum optimizations, 86xx > > > consolidation, e5500/e6500 cpu hotplug, more fman and other dt bits, and > > > minor fixes/cleanup. > > > > Hi Scott, > > > > This one's giving me a few troubles. > > > > > chenhui zhao (6): > > > powerpc/mm: any thread in one core can be the first to setup TLB1 > > > powerpc/cache: add cache flush operation for various e500 > > > powerpc/rcpm: add RCPM driver > > > > That commit doesn't build with SMP=n, as reported by Guenter: > > > > arch/powerpc/sysdev/fsl_rcpm.c:32:2: error: implicit declaration of > > function 'get_hard_smp_processor_id' [-Werror=implicit-function-declaration] > > > > That's not a biggy, I can probably fix it up here. > > By adding asm/smp.h? Yeah I guess. I got derailed today and haven't got around to trying a fix. > I missed this since we don't have a non-SMP defconfig that includes this. Yep, same. I'll add a config to kisskb to catch it in future. > > > powerpc/mpc85xx: Add CPU hotplug support for E6500 > > > > And this one breaks with old binutils (<= 2.22) because of: > > > > arch/powerpc/kernel/head_64.S:210: Error: Unrecognized opcode: `mttmr' > > > > If you're happy to break the build on older bintuils for those configs then > > that's OK with me. > > I'm not too bothered, but it's also an easy fix -- patch sent. Thanks. Do you want to add it to your branch, or I can apply it after I merge your branch? > > > Christophe Leroy (31): > > > powerpc/8xx: Handle CPU6 ERRATA directly in mtspr() macro > > > > This breaks mpc866_ads_defconfig for me, with lots of: > > > > arch/powerpc/mm/8xx_mmu.c:139:2: error: memory input 1 is not directly > > addressable > > > > That's using gcc < 4.9. With >= 4.9 it builds OK. So again your call on > > whether > > we drop support for those compilers for that config. > > I'd rather not require something that new, especially for a chip so old. > > I've sent an untested patch that should fix it. Thanks. I'll wait for Christophe to reply. cheers
Le 15/03/2016 08:01, Scott Wood a écrit : > On Tue, 2016-03-15 at 11:19 +1100, Michael Ellerman wrote: >> On Fri, 2016-03-11 at 21:15 -0600, Scott Wood wrote: >> >>> Highlights include 8xx optimizations, 32-bit checksum optimizations, 86xx >>> consolidation, e5500/e6500 cpu hotplug, more fman and other dt bits, and >>> minor fixes/cleanup. >> Hi Scott, >> >> This one's giving me a few troubles. [...] >>> Christophe Leroy (31): >>> powerpc/8xx: Handle CPU6 ERRATA directly in mtspr() macro >> This breaks mpc866_ads_defconfig for me, with lots of: >> >> arch/powerpc/mm/8xx_mmu.c:139:2: error: memory input 1 is not directly >> addressable >> >> That's using gcc < 4.9. With >= 4.9 it builds OK. So again your call on >> whether >> we drop support for those compilers for that config. > I'd rather not require something that new, especially for a chip so old. > > I've sent an untested patch that should fix it. The proposed patch gives a code slightly different than what Freescale's ERRATA recommends. However, why do we activate CPU6 ERRATA in that config ? CPU6 seems to be an issue with the old MPC860 and MPC855 only. CPU6 is not listed in the MPC866 ERRATAs, so we should probably not have this selected in mpc866_ads_defconfig Christophe PS: It also builds ok with GCC 4.8.3
On Tue, 2016-03-15 at 20:07 +1100, Michael Ellerman wrote: > On Tue, 2016-03-15 at 02:01 -0500, Scott Wood wrote: > > On Tue, 2016-03-15 at 11:19 +1100, Michael Ellerman wrote: > > > On Fri, 2016-03-11 at 21:15 -0600, Scott Wood wrote: > > > > powerpc/mpc85xx: Add CPU hotplug support for E6500 > > > > > > And this one breaks with old binutils (<= 2.22) because of: > > > > > > arch/powerpc/kernel/head_64.S:210: Error: Unrecognized opcode: `mttmr' > > > > > > If you're happy to break the build on older bintuils for those configs > > > then > > > that's OK with me. > > > > I'm not too bothered, but it's also an easy fix -- patch sent. > > Thanks. Do you want to add it to your branch, or I can apply it after I > merge > your branch? If you want to just apply them that'd be fine. -Scott