From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:35524) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1e7EYe-0000sv-87 for qemu-devel@nongnu.org; Wed, 25 Oct 2017 01:48:53 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1e7EYa-0004Of-BB for qemu-devel@nongnu.org; Wed, 25 Oct 2017 01:48:52 -0400 Received: from mx1.redhat.com ([209.132.183.28]:36932) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1e7EYa-0004Nr-57 for qemu-devel@nongnu.org; Wed, 25 Oct 2017 01:48:48 -0400 Received: from smtp.corp.redhat.com (int-mx04.intmail.prod.int.phx2.redhat.com [10.5.11.14]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 3630883F3E for ; Wed, 25 Oct 2017 05:48:47 +0000 (UTC) From: Juan Quintela Date: Wed, 25 Oct 2017 07:48:37 +0200 Message-Id: <20171025054842.24883-1-quintela@redhat.com> Subject: [Qemu-devel] [PATCH v3 0/5] Make xbzrle_cache_size a migration parameter List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: qemu-devel@nongnu.org Cc: dgilbert@redhat.com, lvivier@redhat.com, peterx@redhat.com Hi On v3: - rebase on top of last migration pull requset - improve comments (dave suggestion) - always return 0 in case of error - checked with ovirt/openstack/libvirt people that making value a power of two is not a problem. - should I drop the last patch, or sent it? Please revieww. [v2] - rebase - fix comments about power of 2 instead of multiple of two - minor typo on error message - use Error * when setting values After asking people of libvirt/ovirt/openstack, it appears that there is no problem about restricting cache_size to real values. We give back one error and they had to handle errors already. Please review, Juan. [v1] I started looking at xbzrle_cache_size and saw that it was not a parameter. How difficult could it be? HaHa - x-multifd-page-count and x-multifd-channels were not o migrate_test_apply - page_cache.c has parameters that were not used (max_item_age) - page cache types were random. Move almost all to int/size_t as appropiate - error handling was split in three functions: xbzrle_cache_resize cache_init qmp_set_cache_size Make some sense of it and use Error* - We pass one value to qmp_migrate_cache_size it asigned pow2log() of that value, and cache_init() used a power of two. Make sure that the value passed is right. - Now I am able to make xbzrle_cache_size a parameter!!!! - Now that I have arrived here, I tried to make the json types for migartion parameters the real ones instead of everything is an int64. I will be happy if libvirt people checked this. Please review, Juan. *** BLURB HERE *** Juan Quintela (5): migration: Make sure that we pass the right cache size migration: Don't play games with the requested cache size migration: No need to return the size of the cache migration: Make xbzrle_cache_size a migration parameter migration: [RFC] Use proper types in json hmp.c | 34 ++++++++++++------ migration/migration.c | 94 ++++++++++++++++++++++++++++---------------------- migration/migration.h | 1 - migration/page_cache.c | 12 ++++--- migration/ram.c | 20 ++++------- migration/ram.h | 2 +- qapi/migration.json | 44 ++++++++++++++++------- 7 files changed, 122 insertions(+), 85 deletions(-) -- 2.13.6