[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH] tcg/ppc: disable atomic write check on ppc32
From: |
Richard Henderson |
Subject: |
Re: [Qemu-devel] [PATCH] tcg/ppc: disable atomic write check on ppc32 |
Date: |
Mon, 11 Sep 2017 21:23:25 -0700 |
User-agent: |
Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.3.0 |
On 09/11/2017 02:37 PM, Peter Maydell wrote:
> On 11 September 2017 at 21:49, Philippe Mathieu-Daudé <address@hidden> wrote:
>> this fixes building for ppc64 on ppc32 (changed in 5964fca8a12c):
>>
>> qemu/tcg/ppc/tcg-target.inc.c: In function 'tb_target_set_jmp_target':
>> qemu/include/qemu/compiler.h:86:30: error: static assertion failed: "not
>> expecting: sizeof(*(uint64_t *)jmp_addr) > ATOMIC_REG_SIZE"
>> QEMU_BUILD_BUG_ON(sizeof(*ptr) > ATOMIC_REG_SIZE); \
>> ^
>> qemu/tcg/ppc/tcg-target.inc.c:1377:9: note: in expansion of macro
>> 'atomic_set'
>> atomic_set((uint64_t *)jmp_addr, pair);
>> ^
>>
>> Suggested-by: Richard Henderson <address@hidden>
>> Signed-off-by: Philippe Mathieu-Daudé <address@hidden>
>> ---
>> This fixes Shippable builds, see:
>> https://app.shippable.com/github/qemu/qemu/runs/434/10/console
>>
>> tcg/ppc/tcg-target.inc.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/tcg/ppc/tcg-target.inc.c b/tcg/ppc/tcg-target.inc.c
>> index 21d764c102..0417901289 100644
>> --- a/tcg/ppc/tcg-target.inc.c
>> +++ b/tcg/ppc/tcg-target.inc.c
>> @@ -1374,7 +1374,7 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr,
>> uintptr_t jmp_addr,
>> pair = (uint64_t)i2 << 32 | i1;
>> #endif
>>
>> - atomic_set((uint64_t *)jmp_addr, pair);
>> + atomic_set__nocheck((uint64_t *)jmp_addr, pair);
>> flush_icache_range(jmp_addr, jmp_addr + 8);
>> } else {
>> intptr_t diff = addr - jmp_addr;
>
> Can you explain why this is the right thing? On the
> face of it it looks correct to insist that we don't
> try to do an atomic set of something that's bigger
> than the host can actually handle...
It is the correct thing because ppc32 is handled earlier in the function; only
ppc64 can reach here, therefore a 64-bit atomic_set is always available.
However, I wrote the function intending to minimize the ifdefs so that we can
be sure that it all compiles -- especially the ppc32 bits which I cannot test
on gcc cfarm machines. I didn't think about the fact that ppc32 could not
compile the _Static_assert within the 64-bit atomic_set here in the ppc64
section.
r~