Re: [PATCH v16 00/23] i386 cleanup PART 2

[Claudio Fontana <cfontana@suse.de>]

Hi Philippe,

On 3/14/21 1:00 AM, Philippe Mathieu-Daudé wrote:
> Hi Claudio,
> 
> On 2/4/21 5:39 PM, Claudio Fontana wrote:
>> v15 -> v16:
>>
>> * cpu: Move synchronize_from_tb() to tcg_ops:
>>   - adjusted comments (Alex)
>>
>> * cpu: tcg_ops: move to tcg-cpu-ops.h, keep a pointer in CPUClass:
>>   - remove forward decl. of AccelCPUClass, should be in a later patch. (Alex)
>>   - simplified comment about tcg_ops in struct CPUClass (Alex)
>>   - remove obsolete comment about ARM blocking TCGCPUOps from being const.
>>     (Alex)
>>
>> * accel: replace struct CpusAccel with AccelOpsClass:
>>   - reworded commit message to be clearer about the objective (Alex)
>>
>> * accel: introduce AccelCPUClass extending CPUClass
>>   - reworded commit message to be clearer about the objective (Alex)
>>
>> * hw/core/cpu: call qemu_init_vcpu in cpu_common_realizefn:
>>   - dropped this patch (Alex, Philippe)
>>
>>   will try again later, also in the context of:
>>   https://www.mail-archive.com/qemu-devel@nongnu.org/msg686480.html
>>
>> * accel: introduce new accessor functions
>>   - squashed comments in previous patch introducing accel-cpu.h. (Philippe)
>>
>> * accel-cpu: make cpu_realizefn return a bool
>>   - split in two patches, separating the change to the phys_bits check
>>     (Philippe)
> 
> I am looking at this code:
> 
> $ git grep tcg_ softmmu/physmem.c
> softmmu/physmem.c:153:static void
> tcg_log_global_after_sync(MemoryListener *listener);
> softmmu/physmem.c:154:static void tcg_commit(MemoryListener *listener);
> softmmu/physmem.c:161: * @tcg_as_listener: listener for tracking changes
> to the AddressSpace
> softmmu/physmem.c:167:    MemoryListener tcg_as_listener;
> softmmu/physmem.c:590:static void tcg_iommu_unmap_notify(IOMMUNotifier
> *n, IOMMUTLBEntry *iotlb)
> softmmu/physmem.c:606:static void tcg_register_iommu_notifier(CPUState *cpu,
> softmmu/physmem.c:640:                            tcg_iommu_unmap_notify,
> softmmu/physmem.c:654:void tcg_iommu_free_notifier_list(CPUState *cpu)
> softmmu/physmem.c:668:void tcg_iommu_init_notifier_list(CPUState *cpu)
> softmmu/physmem.c:698:        tcg_register_iommu_notifier(cpu, iommu_mr,
> iommu_idx);
> softmmu/physmem.c:761:    if (tcg_enabled()) {
> softmmu/physmem.c:762:
> newas->tcg_as_listener.log_global_after_sync = tcg_log_global_after_sync;
> softmmu/physmem.c:763:        newas->tcg_as_listener.commit = tcg_commit;
> softmmu/physmem.c:764:
> memory_listener_register(&newas->tcg_as_listener, as);
> softmmu/physmem.c:891:    assert(tcg_enabled());
> softmmu/physmem.c:904:    if (cc->tcg_ops->adjust_watchpoint_address) {
> softmmu/physmem.c:906:        addr =
> cc->tcg_ops->adjust_watchpoint_address(cpu, addr, len);
> softmmu/physmem.c:927:                if (wp->flags & BP_CPU &&
> cc->tcg_ops->debug_check_watchpoint &&
> softmmu/physmem.c:928:
> !cc->tcg_ops->debug_check_watchpoint(cpu, wp)) {
> softmmu/physmem.c:1004:    assert(tcg_enabled());
> softmmu/physmem.c:1059:    if (dirty && tcg_enabled()) {
> softmmu/physmem.c:1107:    if (tcg_enabled()) {
> softmmu/physmem.c:2605:static void
> tcg_log_global_after_sync(MemoryListener *listener)
> softmmu/physmem.c:2634:        cpuas = container_of(listener,
> CPUAddressSpace, tcg_as_listener);
> softmmu/physmem.c:2639:static void tcg_commit(MemoryListener *listener)
> softmmu/physmem.c:2644:    assert(tcg_enabled());
> softmmu/physmem.c:2647:    cpuas = container_of(listener,
> CPUAddressSpace, tcg_as_listener);
> softmmu/physmem.c:2700:        assert(tcg_enabled());
> softmmu/physmem.c:3000:    if (tcg_enabled()) {
> 
> which reminded me the starter generic part of your effort
> (already merged).
> 
> Do you have plans for this code?
> 
> Similarly:
> 
> $ git grep kvm_ softmmu/physmem.c
> softmmu/physmem.c:752:    assert(asidx == 0 || !kvm_enabled());
> softmmu/physmem.c:1295:    if (kvm_enabled())
> softmmu/physmem.c:1296:        kvm_flush_coalesced_mmio_buffer();
> softmmu/physmem.c:1566:    if (kvm_enabled()) {
> softmmu/physmem.c:2046:    if (kvm_enabled() && !kvm_has_sync_mmu()) {
> 
> Thanks,
> 
> Phil.
> 

Hi Phil,

indeed it is a juicy target for splitting things between TCG-only and non-TCG code,
specifically as we discovered that we don't need any of the watchpoint stuff outside of TCG.

I think I am tied up in the ARM code for a while,
so if you are asking because you want to start there, just go ahead,
I'll try to review, otherwise I'll get back to it (and to i386) later on.

Ciao,

Claudio