svn commit: r308130 - in head: . gnu/lib/libgcc gnu/usr.bin/binutils gnu/usr.bin/binutils/ld gnu/usr.bin/binutils/libbfd gnu/usr.bin/cc gnu/usr.bin/gdb gnu/usr.bin/gdb/libgdb lib/libc lib/libc/mips...

[Adrian Chadd]

The places where we set MIPS_SR_COP_1_BIT in locore.S, exception.S.



-adrian


On 10 November 2016 at 13:34, Ruslan Bukin <ruslan.bukin at cl.cam.ac.uk> wrote:
> Hi, Adrian,
>
> thanks, but I have lack of ideas yet.
>
> I just tested this on Ingenic X1000 (mips.mipsel) and it works fine,
> as well as mips.mipselhf (this hardware has FPU).
>
> I have no hardware without FPU.
>
> What exactly COP_1 bits you have commented to get rid of COP1_UNUSABLE?
>
> Ruslan
>
> On Thu, Nov 10, 2016 at 01:09:20PM -0800, Adrian Chadd wrote:
>> ok, so there are two problems with mips now.
>>
>> * COP1_UNUSABLE - which is now being logged, and we were likely
>> triggering it before. The problem: we always turned it on. If I
>> comment out flipping on the COP1 bits in the .S files, those
>> exceptions go away.
>>
>> * Page faults, but that may be me with an older userland:
>>
>> BAD_PAGE_FAULT: pid 1 tid 100001 (init), uid 0: pc 0x404237c0 got a
>> read fault (type 0x2) at 0x8
>> Trapframe Register Dump:
>>     zero: 0    at: 0xffffffffffffffe0    v0: 0x404471b8    v1:
>> 0xffffffffffffffff
>>     a0: 0x404471b8    a1: 0x7ffeea50    a2: 0x16c    a3: 0
>>     t0: 0    t1: 0    t2: 0    t3: 0x70000011
>>     t4: 0x11    t5: 0x70000013    t6: 0x3    t7: 0
>>     t8: 0    t9: 0x4041c940    s0: 0    s1: 0x40417000
>>     s2: 0x7fffeee0    s3: 0xffffffffffffffff    s4: 0xc    s5: 0x40427000
>>     s6: 0x7fffeecc    s7: 0x7ffeea50    k0: 0    k1: 0
>>     gp: 0x4044e630    sp: 0x7ffee838    s8: 0x2    ra: 0x404224d4
>>     sr: 0xfc13    mullo: 0    mulhi: 0    badvaddr: 0x8
>>     cause: 0x8    pc: 0x404237c0
>> Page table info for pc address 0x404237c0: pde = 0x809aa000, pte = 0xa001ba9a
>> Dumping 4 words starting at pc address 0x404237c0:
>> 8e100008 1600fff1 8f828070 100000a3
>>
>>
>>
>> -adrian
>>
>>
>> On 10 November 2016 at 12:49, Adrian Chadd <adrian.chadd at gmail.com> wrote:
>> > hi,
>> >
>> > This fails to boot on actual mips24k hardware, no hardfloat:
>> >
>> > COP1_UNUSABLE: pid 1 tid 100001 (init), uid 0: pc 0x404237c0 ra 0x404224d4
>> > Trapframe Register Dump:
>> >     zero: 0    at: 0xffffffffffffffe0    v0: 0x404471b8    v1:
>> > 0xffffffffffffffff
>> >
>> > .. lots of that the moment we boot userland.
>> >
>> >
>> >
>> > -adrian
>> >
>> >
>> > On 3 November 2016 at 13:10, Ruslan Bukin <ruslan.bukin at cl.cam.ac.uk> wrote:
>> >> On Thu, Nov 03, 2016 at 12:39:48PM -0700, Bryan Drewery wrote:
>> >>> On 10/31/16 8:33 AM, Ruslan Bukin wrote:
>> >>> > Modified: head/share/mk/bsd.cpu.mk
>> >>> > ==============================================================================
>> >>> > --- head/share/mk/bsd.cpu.mk        Mon Oct 31 15:11:55 2016        (r308129)
>> >>> > +++ head/share/mk/bsd.cpu.mk        Mon Oct 31 15:33:58 2016        (r308130)
>> >>> > @@ -303,6 +303,9 @@ MACHINE_CPU = v9 ultrasparc ultrasparc3
>> >>> >
>> >>> >  .if ${MACHINE_CPUARCH} == "mips"
>> >>> >  CFLAGS += -G0
>> >>> > +.if ${TARGET_ARCH:Mmips*hf}
>> >>>
>> >>> TARGET_ARCH is not valid here.  This broke building ports.  Fixed in
>> >>> r308262.
>> >>>
>> >>
>> >> Thank you!
>> >>
>> >> Ruslan
>> >>