On 28.08.19 23:28, Richard Henderson wrote:
> On 8/26/19 1:41 AM, David Hildenbrand wrote:
>>> - /* Make accesses to pages with watchpoints go via the
>>> - watchpoint trap routines. */
>>> - QTAILQ_FOREACH(wp, &cpu->watchpoints, entry) {
>>> - if (cpu_watchpoint_address_matches(wp, vaddr, TARGET_PAGE_SIZE)) {
>>> - /* Avoid trapping reads of pages with a write breakpoint. */
>>> - if ((prot & PAGE_WRITE) || (wp->flags & BP_MEM_READ)) {
>>> - iotlb = PHYS_SECTION_WATCH + paddr;
>>> - *address |= TLB_MMIO;
>>> - break;
>> In the old code, we were able to break once we found a hit ...
>>
>>> - }
>>> - }
>>> + /* Avoid trapping reads of pages with a write breakpoint. */
>>> + match = (prot & PAGE_READ ? BP_MEM_READ : 0)
>>> + | (prot & PAGE_WRITE ? BP_MEM_WRITE : 0);
>>> + flags = cpu_watchpoint_address_matches(cpu, vaddr, TARGET_PAGE_SIZE);
>>> + if (flags & match) {
>> ... now you cannot break early anymore. Maybe pass in the match to
>> cpu_watchpoint_address_matches() ?
>
> Hmm, yes, perhaps.
>
> OTOH, summing a bitmask is a very quick operation. Depending on the total
> number of watchpoints, of course...
And for anything that is not a hit, we have to walk all watchpoints
either way, so the speedup would most probably be neglectable.
--
Thanks,
David / dhildenb
