Re: booke206_tlb_ways() returning 0

[Laurent Vivier]

Le 26/03/2020 à 15:22, Peter Maydell a écrit :
> Hi; Coverity points out a possible issue in booke206_get_tlbm()
> (this is CID 1421942):
> 
> 
> static inline ppcmas_tlb_t *booke206_get_tlbm(CPUPPCState *env, const int 
> tlbn,
>                                               target_ulong ea, int way)
> {
>     int r;
>     uint32_t ways = booke206_tlb_ways(env, tlbn);
>     int ways_bits = ctz32(ways);
>     int tlb_bits = ctz32(booke206_tlb_size(env, tlbn));
>     int i;
> 
>     way &= ways - 1;
>     ea >>= MAS2_EPN_SHIFT;
>     ea &= (1 << (tlb_bits - ways_bits)) - 1;
>     r = (ea << ways_bits) | way;
> 
> Here if booke206_tlb_ways() returns zero, then ways_bits()
> will be 32 and the shift left 'ea << ways_bits' is undefined
> behaviour.
> 
> My first assumption was that booke206_tlb_ways() is not supposed
> to ever return 0 (it's looking at what I think are read-only
> system registers, and it doesn't make much sense to have
> a zero-way TLB). So I tried adding an assert:
> 
> diff --git a/target/ppc/cpu.h b/target/ppc/cpu.h
> index 88d94495554..aedb6bcb265 100644
> --- a/target/ppc/cpu.h
> +++ b/target/ppc/cpu.h
> @@ -2403,6 +2403,7 @@ static inline int booke206_tlb_ways(CPUPPCState
> *env, int tlbn)
>  {
>      uint32_t tlbncfg = env->spr[SPR_BOOKE_TLB0CFG + tlbn];
>      int r = tlbncfg >> TLBnCFG_ASSOC_SHIFT;
> +    assert(r > 0);
>      return r;
>  }
> 
> However, this isn't right, because it causes one of the check-acceptance
> tests to fail, with this backtrace:
> 
> #3  0x00007ffff074d412 in __GI___assert_fail (assertion=0x5555560a0d7d
> "r > 0", file=0x5555560a0d40
> "/home/petmay01/linaro/qemu-from-laptop/qemu/target/ppc/cpu.h",
> line=2406, function=0x5555560a1720 <__PRETTY_FUNCTION__.20811>
> "booke206_tlb_ways") at assert.c:101
> #4  0x0000555555a9939b in booke206_tlb_ways (env=0x555556e52a30,
> tlbn=2) at /home/petmay01/linaro/qemu-from-laptop/qemu/target/ppc/cpu.h:2406
> #5  0x0000555555a9b3ac in mmubooke206_get_physical_address
> (env=0x555556e52a30, ctx=0x7fffd77008a0, address=9223380835095282947,
> rw=0, access_type=0, mmu_idx=1) at
> /home/petmay01/linaro/qemu-from-laptop/qemu/target/ppc/mmu_helper.c:1093
> #6  0x0000555555a9c25d in get_physical_address_wtlb
> (env=0x555556e52a30, ctx=0x7fffd77008a0, eaddr=9223380835095282947,
> rw=0, access_type=0, mmu_idx=1) at
> /home/petmay01/linaro/qemu-from-laptop/qemu/target/ppc/mmu_helper.c:1455
> #7  0x0000555555a9c82b in cpu_ppc_handle_mmu_fault
> (env=0x555556e52a30, address=9223380835095282947, rw=0, mmu_idx=1)
>     at 
> /home/petmay01/linaro/qemu-from-laptop/qemu/target/ppc/mmu_helper.c:1597
> #8  0x0000555555a9f975 in ppc_cpu_tlb_fill (cs=0x555556e49560,
> addr=9223380835095282947, size=1, access_type=MMU_DATA_LOAD,
> mmu_idx=1, probe=false, retaddr=140735610345781) at
> /home/petmay01/linaro/qemu-from-laptop/qemu/target/ppc/mmu_helper.c:3053
> #9  0x00005555558e1422 in tlb_fill (cpu=0x555556e49560,
> addr=9223380835095282947, size=1, access_type=MMU_DATA_LOAD,
> mmu_idx=1, retaddr=140735610345781) at
> /home/petmay01/linaro/qemu-from-laptop/qemu/accel/tcg/cputlb.c:1017
> #10 0x00005555558e279b in load_helper (env=0x555556e52a30,
> addr=9223380835095282947, oi=1, retaddr=140735610345781, op=MO_8,
> code_read=false, full_load=0x5555558e2b3a <full_ldub_mmu>) at
> /home/petmay01/linaro/qemu-from-laptop/qemu/accel/tcg/cputlb.c:1534
> #11 0x00005555558e2b80 in full_ldub_mmu (env=0x555556e52a30,
> addr=9223380835095282947, oi=1, retaddr=140735610345781)
>     at /home/petmay01/linaro/qemu-from-laptop/qemu/accel/tcg/cputlb.c:1624
> #12 0x00005555558e2bb4 in helper_ret_ldub_mmu (env=0x555556e52a30,
> addr=9223380835095282947, oi=1, retaddr=140735610345781)
>     at /home/petmay01/linaro/qemu-from-laptop/qemu/accel/tcg/cputlb.c:1630
> #13 0x00007fff900fd9c6 in code_gen_buffer ()
> #14 0x00005555558f9915 in cpu_tb_exec (cpu=0x555556e49560,
> itb=0x7fff900fd780 <code_gen_buffer+1038163>)
>     at /home/petmay01/linaro/qemu-from-laptop/qemu/accel/tcg/cpu-exec.c:172
> #15 0x00005555558fa732 in cpu_loop_exec_tb (cpu=0x555556e49560,
> tb=0x7fff900fd780 <code_gen_buffer+1038163>, last_tb=0x7fffd7701078,
> tb_exit=0x7fffd7701070) at
> /home/petmay01/linaro/qemu-from-laptop/qemu/accel/tcg/cpu-exec.c:619
> #16 0x00005555558faa4c in cpu_exec (cpu=0x555556e49560) at
> /home/petmay01/linaro/qemu-from-laptop/qemu/accel/tcg/cpu-exec.c:732
> #17 0x00005555558bcf29 in tcg_cpu_exec (cpu=0x555556e49560) at
> /home/petmay01/linaro/qemu-from-laptop/qemu/cpus.c:1405
> #18 0x00005555558bd77f in qemu_tcg_cpu_thread_fn (arg=0x555556e49560)
> at /home/petmay01/linaro/qemu-from-laptop/qemu/cpus.c:1713
> #19 0x0000555555f2ff3f in qemu_thread_start (args=0x555556e8dd10) at
> /home/petmay01/linaro/qemu-from-laptop/qemu/util/qemu-thread-posix.c:519
> #20 0x00007ffff0b156db in start_thread (arg=0x7fffd7704700) at
> pthread_create.c:463
> #21 0x00007ffff083e88f in clone () at
> ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
> 
> So under what circumstances can booke206_tlb_ways()
> validly return 0? Should booke206_get_tlbm() cope with a
> zero return, or can it assert() that it will never
> call booke206_tlb_ways() in a way that will cause one?

It seems the value of booke206_tlb_ways() is the associativity
(TLBnCFG_ASSOC_SHIFT) that seems to be generated by gen_tlbncfg() and
set in init_proc_e500().

And used values are: 2, 4 ,16, and 64, but only for tlbn 0 and 1.

According to PowerISA 2.06:

booke206_tlb_size() booke206_tlb_ways()

0                   0           No TLB present
0                   1           TLB geometry is completely
                                implementation-defined.
                                MAS0 ESEL is ignored
0                   > 1         TLB geometry and number of
                                entries is implementation
                                defined,...
n > 0               n or 0      TLB is fully associative

In mmubooke206_get_physical_address(), helper_booke206_tlbsx(),
booke206_invalidate_ea_tlb() and helper_booke206_tlbilx3(),
booke206_get_tlbm() is not called if ways is 0.

In booke206_cur_tlb(), the function is only called  with tlbn number
provided by SPR_BOOKE_MAS0, so we can guess we are using the ones with
the initialized values.

It is also called in mmubooke_create_initial_mapping() but only for tlbn 1.

But r can be 0 if ea and way are 0 (in
mmubooke206_get_physical_address() if ways > 0, way will start with
value 0 and ea >> ways_bit can be 0). I think it's why your assert() is
triggered.

Thanks,
Laurent