xbps_file_hash: reduce memory footprint and process file in small chunks.
With xbps-rindex(8) -c/-r being multithreaded, the memory usage was insanely high in the build server, with peaks of almost 4GB (this depends on the file size being processed and number of threads). Process the file in small chunks instead to compute the SHA256 hash.
This commit is contained in:
4
NEWS
4
NEWS
@@ -1,5 +1,9 @@
|
||||
xbps-0.40 (???):
|
||||
|
||||
* Reduced considerably memory requirement when checking SHA256 hashes on
|
||||
files. Previous way of allocating heap memory as big as the file being
|
||||
processed wasn't a great idea.
|
||||
|
||||
* xbps-{install,remove}: the post-install/pre-remove messages are now
|
||||
printed with some guards, i.e:
|
||||
|
||||
|
Reference in New Issue
Block a user