ligang/vpp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

vhost: Fix mmap size calculation

Browse Source 
I had a bug where a requested size of 1G was resulting in
an aligned size of '1G + 2M', resulting in an OOM error.

Previous code was adding one huge page size
when memory is already aligned.

Change-Id: Idd3aa0e9b893fb3efccba6ae1c7161e26d3f9456
Signed-off-by: Pierre Pfister <ppfister@cisco.com>
This commit is contained in:
Pierre Pfister
2017-04-20 15:34:00 +02:00
committed by Damjan Marion
parent 1985c93bd7
commit c4352553ee
1 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/vnet/devices/virtio/vhost-user.c
View File

@ -303,7 +303,7 @@ unmap_all_mem_regions (vhost_user_intf_t * vui)
ssize_t map_sz = (vui->regions[i].memory_size +
vui->regions[i].mmap_offset +
page_sz) & ~(page_sz - 1);
page_sz - 1) & ~(page_sz - 1);
r =
munmap (vui->region_mmap_addr[i] - vui->regions[i].mmap_offset,
@ -928,7 +928,7 @@ vhost_user_socket_read (unix_file_t * uf)
/* align size to 2M page */
ssize_t map_sz = (vui->regions[i].memory_size +
vui->regions[i].mmap_offset +
page_sz) & ~(page_sz - 1);
page_sz - 1) & ~(page_sz - 1);
vui->region_mmap_addr[i] = mmap (0, map_sz, PROT_READ | PROT_WRITE,
MAP_SHARED, fds[i], 0);
@ -1179,7 +1179,7 @@ vhost_user_socket_read (unix_file_t * uf)
/* align size to 2M page */
long page_sz = get_huge_page_size (fd);
ssize_t map_sz =
(msg.log.size + msg.log.offset + page_sz) & ~(page_sz - 1);
(msg.log.size + msg.log.offset + page_sz - 1) & ~(page_sz - 1);
vui->log_base_addr = mmap (0, map_sz, PROT_READ | PROT_WRITE,
MAP_SHARED, fd, 0);