While working on adding features to the "tomato" firmware distro for my WRT54G linksys based router (which uses busybox 1.12.3) I came across some problems trying to implement sqlite to store network bandwidth statistics on a cifs mounted share. I noticed the same behaviour as mentioned in these links: http://www.nabble.com/Bug-483502:-smbfs:-mount--o-remount,-brl-or--o-remount,-nobrl-changes-the-reported-%22mand%22-option,-but-has-no-effect-td17527313.html http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=483216 When sqlite started failing 100% I started digging and found many postings around the net where people were having the same problems. There is a simple test program in the first link mentioned above which I have confirmed fails 100% on my box in the exact same location mentioned in that link (during locking write's) on a cifs mounted system with -EPERM (-1). Running the same problem on the local busybox system the same test does not fail.
unless i missed something, i dont see how this is a busybox problem at all
Sorry I am totally new to embedded linux, cross compilers etc.. so I didn't quite understand where busybox, uclib, etc.. all fit into the big picture. Now I do. There is a SAMBA server defect in the distro I am using (which happens to run on busybox) and I assumed that the mount.c (being part of busybox) had the defect but alas it was not busybox at all. Sorry about that. Apparently the defect is fixed AFTER samba server 3.0.22 (ie: 3.0.33 and beyond). Thanks