Browse Source

Use dumpe2fs for filesystem size calculations

The previous approach used `df` to get usable space and then added a
fixed size to that number in order to account for filesystem
overhead. However, at some point that stopped working for me. It
appears that ext4 filesystem overhead can vary over time or because of
other factors. (Certainly now that I think about it the old code would
only work well for people with the exact same filesystem size as me.)

So the new approach is to just completely ignore what `df` tells us
and instead go directly to the source: the filesystem's internal
notion of exactly how much space it takes up. We use `dumpe2fs` to
retrieve this information and calculate the on-disk size dynamically
from that. Then we add the space that boot data takes up (unchanged),
and we add 5MB padding because when I tested this it didn't quite add
up otherwise. https://unix.stackexchange.com/a/13551/29146 suggests
that this unaccounted-for data may be e.g. additional copies of the
superblock.
AJ Jordan 5 years ago
parent
commit
593724c05b
1 changed files with 13 additions and 5 deletions
  1. 13 5
      init/resize-rootfs-if-needed.sh

+ 13 - 5
init/resize-rootfs-if-needed.sh

@@ -14,12 +14,20 @@ fi
 sysfs_xvda="/sys/class/block/xvda"
 
 # if root filesystem is already using (almost) the whole disk
-# 203M for BIOS and /boot data, 222 for ext4 filesystem overhead
-# See QubesOS/qubes-core-agent-linux#146 for more details
-size_margin=$(((222 + 203) * 2 * 1024))
-rootfs_size=$(df --block-size=512 --output=size / | tail -n 1 | tr -d ' ')
+# 203M for BIOS and /boot data
+boot_data_size=$((203 * 2 * 1024))
+# rootfs size is calculated on-the-fly. `df` doesn't work because it doesn't
+# include fs overhead, and calculating a static size for overhead doesn't work
+# because that can change dynamically over the filesystem's lifetime.
+# See QubesOS/qubes-core-agent-linux#146 and QubesOS/qubes-core-agent-linux#152
+# for more details
+ext4_block_count=$(dumpe2fs /dev/mapper/dmroot | grep '^Block count:' | sed -E 's/Block count:[[:space:]]+//')
+ext4_block_size=$(dumpe2fs /dev/mapper/dmroot | grep '^Block size:' | sed -E 's/Block size:[[:space:]]+//')
+rootfs_size=$((ext4_block_count * ext4_block_size / 512))
+# 5 MB in 512-byte units for some random extra bits
+size_margin=$((5 * 1024 * 2))
 if [ "$(cat $sysfs_xvda/size)" -lt \
-       $(( size_margin + rootfs_size )) ]; then
+       $(( rootfs_size + boot_data_size + size_margin )) ]; then
    echo "root filesystem already at $rootfs_size blocks" >&2
    exit 0
 fi