Progress bar updates flicker quite a bit on slow hw / high resolutions
as the background is completely cleared before the new progress bar
position is drawn on top.
Improve it by first drawing the progress bar and then only fill the
remaining rows with the background.
function old new delta
fb_drawprogressbar 444 429 -15
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
The console passed to TIOCCONS has to be writable, otherwise future
console writes will fail.
This presumably used to work, but in current kernels (see
drivers/tty/tty_io.c:redirected_tty_write) console writes are sent to
vfs_write(device), which checks if the device is writable. A quick look
in the linux git history doesn't show any recent changes to either tty_io
or vfs_write.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
If CONFIG_FEATURE_HTTPD_GZIP is enabled and request contained
'Accept-Encoding: gzip', then errors were sent with 'Content-Encoding: gzip'
even though they aren't.
Fix it by clearing content_gzip before sending the headers.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Don't fail on unknown nfs mount options. The kernel automatically
adds a few mount options that we don't currently handle, like:
sec=null|sys|krb5..
mountaddr=<ip>
mountproto=tcp|udp
local_lock=none|all|flock|posix
Which causes a simple mount -o remount,rw <mountpoint> to fail.
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
function old new delta
handle_incoming_and_exit 2830 2807 -23
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Peter Korsgaard