Re: [Qemu-devel] [PATCH v3 27/32] s390x/tcg: Provide probe_write_access helper

[David Hildenbrand <david@redhat.com>]

On 07.03.19 15:10, Richard Henderson wrote:
> On 3/7/19 4:15 AM, David Hildenbrand wrote:
>> +void probe_write_access(CPUS390XState *env, uint64_t addr, uint64_t len,
>> +                        uintptr_t ra)
>> +{
>> +#ifdef CONFIG_USER_ONLY
>> +    if (!h2g_valid(addr) || !h2g_valid(addr + len - 1)) {
>> +        s390_program_interrupt(env, PGM_ADDRESSING, ILEN_AUTO, ra);
>> +    }
> 
> You need
> 
>   || page_check_range(addr, len, PAGE_WRITE) < 0
> 
> as well.

Indeed, thanks.

So it should be


+void probe_write_access(CPUS390XState *env, uint64_t addr, uint64_t len,
+                        uintptr_t ra)
+{
+#ifdef CONFIG_USER_ONLY
+    if (!h2g_valid(addr) || !h2g_valid(addr + len - 1) ||
+        page_check_range(addr, len, PAGE_WRITE) < 0) {
+        s390_program_interrupt(env, PGM_ADDRESSING, ILEN_AUTO, ra);
+    }
+#else
+    /* test the actual access, not just any access to the page due to LAP */
+    while (len) {
+        const uint64_t pagelen = -(addr | -TARGET_PAGE_MASK);
+        const uint64_t curlen = MIN(pagelen, len);
+
+        probe_write(env, addr, curlen, cpu_mmu_index(env, false), ra);
+        addr = wrap_address(env, addr + curlen);
+        len -= curlen;
+    }
+#endif
+}

Conny, I can resend if you don't feel like fixing up (or there is more to do).

-- 

Thanks,

David / dhildenb