jsalter@ibmpa.awdpa.ibm.com (04/17/91)
In article <1991Apr16.160043.20354@ux1.cso.uiuc.edu> kline@ux1.cso.uiuc.edu (Charley Kline) writes: >jsalter@ibmpa.awdpa.ibm.com writes: >>You must be running on an *early* version (pre-GA?). You need an upgrade. No, you're not. >Gee, I'm on the same system Phil is. It's an S6000 model 540 running 3003. >I can't get df to misbehave either. Oops, apologies. Casual probing didn't uncover your problem, but was brought to my attention that this does happen on the latest, but wasn't easily reproducable. After a little bit of hacking, I am now able to reproduce it: % df . Filesystem Total KB free %used iused %iused Mounted on /dev/lv02 200704 4944 97% 12439 24% /u % chmod 750 . Filesystem Total KB free %used iused %iused Mounted on Cannot find file system . % Sure enough, if the access mode for the directory doesn't have the user's executable bit on, df . doesn't work. A defect has been opened. >Charley Kline, KB9FFK, PP-ASEL c-kline@uiuc.edu >University of Illinois Computing Services Packet: kb9ffk@w9yh >1304 W. Springfield Ave, Urbana IL 61801 (217) 333-3339 jim/jsalter IBM PSP, Palo Alto T465/(415)855-4427 VNET: JSALTER at AUSVMQ Internet: jsalter@slo.awdpa.ibm.com UUCP: ..!uunet!ibmsupt!jsalter "IBM part #23521, aka Lt. Commander Data" The stuff above is on my own.
woan@nowhere (Ronald S Woan) (04/17/91)
In article <1991Apr16.210447.28136@ibmpa.awdpa.ibm.com> jsalter@slo.awdpa.ibm.com (Jim Salter) writes: >Sure enough, if the access mode for the directory doesn't have the user's >executable bit on, df . doesn't work. A defect has been opened. Are you positive that this is a defect? I tried on an old Sun 3 (SunOS 4.??) last night and am pretty sure that I get the same error... -- +-----All Views Expressed Are My Own And Are Not Necessarily Shared By------+ +------------------------------My Employer----------------------------------+ + Ronald S. Woan woan@cactus.org or woan@austin.vnet.ibm.com + + other email addresses Prodigy: XTCR74A Compuserve: 73530,2537 +
shore@theory.tn.cornell.edu (Melinda Shore) (04/18/91)
In article <3783@d75.UUCP> woan@cactus.org writes: >In article <1991Apr16.210447.28136@ibmpa.awdpa.ibm.com> jsalter@slo.awdpa.ibm.com (Jim Salter) writes: >>Sure enough, if the access mode for the directory doesn't have the user's >>executable bit on, df . doesn't work. A defect has been opened. >Are you positive that this is a defect? This is a bug, maybe. In Unix filesystem semantics, if the permissions on a directory are set up so that the relevant (to the user opening the directory) read bits are on but the execute bits are off, the user should be able to read the directory but not any files in the directory. It also means that the user should be able to stat(2) the directory. However, it means that the user should not be able to cd into the directory. Now, if the permissions look something like drwxr--r-- 6 shore staff ... shore will be able to access files in the directory, but nobody else (other than root) will. The only way that I can think of off the top of my head that doing "df ." should fail for a legitimate reason would be if df is setuid someone other than root, and the directory permissions were as peculiar as those listed above. -- Software longa, hardware brevis Melinda Shore - Cornell Information Technologies - shore@theory.tn.cornell.edu
web@farpoint.austin.ibm.com (Bill Baker) (04/18/91)
>In article <3783@d75.UUCP> woan@cactus.org writes: >>In article <1991Apr16.210447.28136@ibmpa.awdpa.ibm.com> jsalter@slo.awdpa.ibm.com (Jim Salter) writes: >>>Sure enough, if the access mode for the directory doesn't have the user's >>>executable bit on, df . doesn't work. A defect has been opened. >>Are you positive that this is a defect? In the old days, df had to grope the super block to determine the free space in a file system. As a result, df had to be set-uid root, set-uid to the owner of the device files, or set-gid to the group id of the device files. Newer versions of Unix have a statfs (or statvfs) system call that returns interesting file system statistics given the name of any file in the file system. That makes "df ." trivial to implement. Df in AIX was coded to use statfs. However, it was mistakenly left set-uid bin. This means that df will fail in any directory that does not give search access to bin. This has already been fixed in AIX V3.1.5. web -- Bill Baker Internet: web@glasnost.austin.ibm.com IBM PSP AWD net: web@farpoint.austin.ibm.com 11400 Burnet Rd. VNET: WEBAKER AT AUSVMQ Austin, TX; 78758-2502