| Commit-Queue | +1 |
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
case 79: // FileIo_V2_QueryFullSizeInfo
case 80: // QuerySetVolumeInfo
break;Should we still log the "unhandled file op code", or it this changing because we intentionally ignore them?
// Size of `FileIo_Info` event: 3 pointers + 4 `uint32`s.In decoding below I see 4 * pointer_size and 3 * uint32_t?
Although in the test, `extra_info` (pointer_size) seems to match operation_status (uint32_t)
#if defined(ARCH_CPU_64_BITS)
static constexpr uint16_t kEventHeaderFlags = EVENT_HEADER_FLAG_64_BIT_HEADER;
#else
static constexpr uint16_t kEventHeaderFlags = EVENT_HEADER_FLAG_32_BIT_HEADER;
#endif
Both should work on either 64bit or 32 bit target really, but I guess it's a good change to at least exercise each.
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
PTAL
case 79: // FileIo_V2_QueryFullSizeInfo
case 80: // QuerySetVolumeInfo
break;Should we still log the "unhandled file op code", or it this changing because we intentionally ignore them?
We previously weren't logging unhandled file op codes - I intended the logging to check for op codes we didn't know about. I've added a TODO to handle the Query op codes. I don't think they're used that often so they shouldn't cause many unmatched end events, but we should handle them for completeness.
// Size of `FileIo_Info` event: 3 pointers + 4 `uint32`s.In decoding below I see 4 * pointer_size and 3 * uint32_t?
Although in the test, `extra_info` (pointer_size) seems to match operation_status (uint32_t)
Good point - Gemini suggested storing the status in the extra_info field. I've switched the status to 32 bits and stuck it in the 64 bit extra_info field.
#if defined(ARCH_CPU_64_BITS)
static constexpr uint16_t kEventHeaderFlags = EVENT_HEADER_FLAG_64_BIT_HEADER;
#else
static constexpr uint16_t kEventHeaderFlags = EVENT_HEADER_FLAG_32_BIT_HEADER;
#endif
Both should work on either 64bit or 32 bit target really, but I guess it's a good change to at least exercise each.
This is just to stop the CHECK_EQ check above from firing.
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
| Code-Review | +1 |
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |
Add Flt file i/o opcodes support for tracing
I've landed Perfetto support for the Flt file i/o opcodes. This CL
adds trace recording for these opcodes.
gemini-cli wrote the additional tests. This CL also fixes the ETW
tests on x86 builds.
| Inspect html for hidden footers to help with email filtering. To unsubscribe visit settings. |