summaryrefslogtreecommitdiffstats
path: root/src/commonlib/include/commonlib/timestamp_serialized.h
diff options
context:
space:
mode:
Diffstat (limited to 'src/commonlib/include/commonlib/timestamp_serialized.h')
-rw-r--r--src/commonlib/include/commonlib/timestamp_serialized.h8
1 files changed, 8 insertions, 0 deletions
diff --git a/src/commonlib/include/commonlib/timestamp_serialized.h b/src/commonlib/include/commonlib/timestamp_serialized.h
index bc2152eab438..4b1001ef3927 100644
--- a/src/commonlib/include/commonlib/timestamp_serialized.h
+++ b/src/commonlib/include/commonlib/timestamp_serialized.h
@@ -199,6 +199,14 @@ static const struct timestamp_id_to_name {
TS_NAME_DEF(TS_BOOTBLOCK_END, 0, "end of bootblock"),
TS_NAME_DEF(TS_COPYROM_START, TS_COPYROM_END, "starting to load romstage"),
TS_NAME_DEF(TS_COPYROM_END, 0, "finished loading romstage"),
+ /*
+ * "ignore for x86": On platforms with memory-mapped flash, it's
+ * impossible to separate loading times from decompression times because
+ * the flash accesses happen in the background as the decompression is
+ * running. So it makes more sense to consider the total loading +
+ * decompression time by looking at the timestamps before and after the
+ * decompression timestamps.
+ */
TS_NAME_DEF(TS_ULZMA_START, TS_ULZMA_END, "starting LZMA decompress (ignore for x86)"),
TS_NAME_DEF(TS_ULZMA_END, 0, "finished LZMA decompress (ignore for x86)"),
TS_NAME_DEF(TS_ULZ4F_START, TS_ULZ4F_END, "starting LZ4 decompress (ignore for x86)"),