On Wed, Sep 3, 2014 at 11:06 PM, Aaro Koskinen <[email protected]> wrote: > On Wed, Sep 03, 2014 at 06:36:36PM +0200, Denys Vlasenko wrote: >> On Mon, Sep 1, 2014 at 10:24 PM, Aaro Koskinen <[email protected]> wrote: >> > The firmware in some HP PA-RISC boxes is always sending fixed >> > 512-byte requests, and sometimes there is some garbage at the end >> > which makes busybox to reject the request. Ignore any characters after >> > the last NUL. >> >> How about this? > > As you pointed out, I missed the short read possibility, and by > ignoring them that also made me to screw up my testing & analysis > (should have checked with tcpdump) - the box is actually sending > *516-byte* requests. :-( > > So following is still needed to current git to work with some PA-RISC: > > diff --git a/networking/tftp.c b/networking/tftp.c > index 8e3b0a2..d9295c9 100644 > --- a/networking/tftp.c > +++ b/networking/tftp.c > @@ -118,7 +118,7 @@ struct globals { > uint8_t error_pkt[4 + 32]; > struct passwd *pw; > /* used in tftpd_main(), a bit big for stack: */ > - char block_buf[TFTP_BLKSIZE_DEFAULT]; > + char block_buf[TFTP_BLKSIZE_DEFAULT + 4]; > char block_buf_tail[1]; > #if ENABLE_FEATURE_TFTP_PROGRESS_BAR > off_t pos; > @@ -811,7 +811,7 @@ int tftpd_main(int argc UNUSED_PARAM, char **argv) > ) { > goto err; > } > - /* Some HP PA-RISC firmware always sends fixed 512-byte requests, > + /* Some HP PA-RISC firmware always sends fixed 516-byte requests, > * with trailing garbage. > * Support that by not requiring NUL to be the last byte (see above). > * To make strXYZ() ops safe, force NUL termination:
Fixed in git, thanks! _______________________________________________ busybox mailing list [email protected] http://lists.busybox.net/mailman/listinfo/busybox
