Afer Zram backing dev setuped, zram will be treated as a ASYNC IO
swap device, no longer SYNC IO device, because WB pages do have IO
operation on swap in, so the optimization on skiping swapcache
for fast swap device will no longer be applyed on such case.
Now we refactor this optimization, skip the swapcache by checking
SYNC IO state on each swap page via ioctl. Reuse ioctl because we
should not break the GKI.
Change-Id: Ie4316c9a43f481ccd4dff6b90c63da108bed916c
Signed-off-by: lulei1 <lulei1@motorola.com>
Reviewed-on: https://gerrit.mot.com/2156790
SLTApproved: Slta Waiver
SME-Granted: SME Approvals Granted
Tested-by: Jira Key
Reviewed-by: Zhangqing Huang <huangzq2@motorola.com>
Reviewed-by: Guolin Wang <wanggl3@lenovo.com>
Reviewed-by: Xiangpo Zhao <zhaoxp3@motorola.com>
Submit-Approved: Jira Key
lulei1
eb80d2507d