file: read in entire file (up to 4K) instead of stat()'ing for size

We use find_string() to check files in both /proc (e.g.,
/proc/device-tree/...) and /sys (e.g., /sys/bus/spi/devices/...).
find_string() currently tries to iteratively read pieces of these files,
using stat() to get the file size. However, this size is inaccurate on
sysfs, as these files are often dynamically generated, so it reports a
size of PAGE_SIZE (i.e., 4096).

Instead of trying to pre-determine the file size, let's read in the
whole file (up to 4KiB - 1) and do a series of string searches to find
any matching strings. The 4KiB limitation should be OK, since we're only
looking for things like modalias and compatible strings.

As a side effect, this eliminates a lot of unnecessary strlen() calls,
and allows us to again avoid hand-rolling strstr().

BUG=chrome-os-partner:51651
BRANCH=none
TEST=`flashrom --wp-status` doesn't show log spam
TEST=`flashrom --read /tmp/flashX` gives same results before/after

Change-Id: I2d053b18ba624f4ce3ebc1956b9e7945b693b0bc
Signed-off-by: Brian Norris <briannorris@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/335353
Reviewed-by: David Hendricks <dhendrix@chromium.org>
---
M file.c
1 file changed, 41 insertions(+), 44 deletions(-)