Fix viz memory dumping

When there are multiple windows open with viz, each window has it's own
InProcCmdBuffer. If we do a memory dump with multiple windows, we will
fail DCHECKs on calls from VizProcessContextProvider::OnMemoryDump
because we will have multiple CommandBufferHelpers in the same process
with the same ring_buffer_id_.

https://cs.chromium.org/chromium/src/gpu/command_buffer/client/cmd_buffer_helper.cc?rcl=9c52bb1d0f58d02879f4373f5db09aa625c590a7&l=365

This CL makes the transfer buffer IDs generated by InProcCmdBuffer
unique within a process and avoids the failing check.

Cq-Include-Trybots: luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
Change-Id: I204098af24c6148c1abe71724dbc5b176ee1a29f
Reviewed-on: https://chromium-review.googlesource.com/1243672Reviewed-by: Antoine Labour <piman@chromium.org>
Commit-Queue: Jonathan Backer <backer@chromium.org>
Cr-Commit-Position: refs/heads/master@{#594084}

gpu/ipc/in_process_command_buffer.cc

@@ -89,6 +89,7 @@ namespace {
 
 base::AtomicSequenceNumber g_next_route_id;
 base::AtomicSequenceNumber g_next_image_id;
+base::AtomicSequenceNumber g_next_transfer_buffer_id;
 
 CommandBufferId NextCommandBufferId() {
   return CommandBufferIdFromChannelAndRoute(kInProcessCommandBufferClientId,
@@ -934,7 +935,7 @@ scoped_refptr<Buffer> InProcessCommandBuffer::CreateTransferBuffer(
     size_t size,
     int32_t* id) {
   scoped_refptr<Buffer> buffer = MakeMemoryBuffer(size);
-  *id = ++next_transfer_buffer_id_;
+  *id = g_next_transfer_buffer_id.GetNext() + 1;
   ScheduleGpuTask(
       base::BindOnce(&InProcessCommandBuffer::RegisterTransferBufferOnGpuThread,
                      gpu_thread_weak_ptr_factory_.GetWeakPtr(), *id, buffer));

gpu/ipc/in_process_command_buffer.h

@@ -330,7 +330,6 @@ class GL_IN_PROCESS_CONTEXT_EXPORT InProcessCommandBuffer
   int32_t last_put_offset_ = -1;
   Capabilities capabilities_;
   GpuMemoryBufferManager* gpu_memory_buffer_manager_ = nullptr;
-  int32_t  next_transfer_buffer_id_ = 1;
   uint64_t next_fence_sync_release_ = 1;
   uint64_t flushed_fence_sync_release_ = 0;
   std::vector<SyncToken> next_flush_sync_token_fences_;