NetBSD Problem Report #7084
Received: (qmail 5073 invoked from network); 5 Mar 1999 03:00:54 -0000
Message-Id: <199903050300.TAA00967@isaac.nas.nasa.gov>
Date: Thu, 4 Mar 1999 19:00:20 -0800 (PST)
From: "Thomas M. Proett" <proett@nas.nasa.gov>
Reply-To: proett@nas.nasa.gov
To: gnats-bugs@gnats.netbsd.org
Subject: read.2 man page
X-Send-Pr-Version: 3.95
>Number: 7084
>Category: lib
>Synopsis: read.2 man page needs entry for EINVAL
>Confidential: no
>Severity: non-critical
>Priority: medium
>Responsible: lib-bug-people
>State: closed
>Class: doc-bug
>Submitter-Id: net
>Arrival-Date: Thu Mar 04 19:05:01 +0000 1999
>Closed-Date: Wed May 17 10:16:46 +0000 2000
>Last-Modified: Wed May 17 10:16:46 +0000 2000
>Originator: Tom Proett
>Release: <NetBSD-current source date> NetBSD 1.3J
>Organization:
proett@nas.nasa.gov
NASA Ames Research Center
>Environment:
System: NetBSD isaac.nas.nasa.gov 1.3J NetBSD 1.3J (MSS3_RAWHIDE) #3: Thu Mar 4 14:45:11 PST 1999 root@isaac.nas.nasa.gov:/usr/src/sys/arch/alpha/compile/MSS3_RAWHIDE alpha
>Description:
A program which opens a raw disk device then tries to read anything
except an even multiple of a block size gets EINVAL. This is not
mentioned in the read.2 man page.
>How-To-Repeat:
Compile and run:
#include <stdlib.h>
#include <fcntl.h>
#include <errno.h>
main()
{
int fd;
int cnt;
char buf[50];
if ((fd = open("/dev/rsd10g", O_RDONLY)) == -1) {
perror("open");
exit(1);
}
cnt = read(fd, buf, sizeof(buf));
printf("got %d with err %d\n", cnt, errno);
return 0;
}
You get:
got -1 with err 22
The line of interest from /usr/include/sys/errno.h is:
#define EINVAL 22 /* Invalid argument */
>Fix:
>Release-Note:
>Audit-Trail:
From: "Erik E. Fair" <fair@clock.org>
To: proett@nas.nasa.gov
Cc: gnats-bugs@gnats.netbsd.org
Subject: Re: lib/7084: read.2 man page
Date: Fri, 5 Mar 1999 09:51:28 -0800
The read.2 doesn't mention the myriad error messages one could get from a
network failure while reading a socket, either. I don't think that is a
bug, however; we need to cross reference stuff more, but while a little
duplication of information in the man pages is OK, it increases the
probability of documentation synchronization errors, and should be avoided
in general.
Perhaps a reference to intro(4) is in order. Hmmm. I can't find the doco
that says that raw devices have to be read/written in DEV_BSIZE chunks...
Erik <fair@clock.org>
State-Changed-From-To: open->closed
State-Changed-By: fair
State-Changed-When: Wed May 17 03:15:07 PDT 2000
State-Changed-Why:
the basic system call man pages are not a catchall for every error that
a descriptor operation can generate, particularly given that a descriptor
can refer to a whole lot of different objects, each of which has its own
characteristics (and, necessarily, error messages associated with its
own peculiar failure modes).
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.51 2026/08/10 02:28:17 riastradh Exp $
$NetBSD: gnats_config.sh,v 1.10 2026/05/13 22:00:09 riastradh Exp $
Copyright © 1994-2026
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.