The newboot rootconf() subroutine
http://cvs.opensolaris.org/source/xref/usr/src/uts/common/fs/vfs.c#rootconf
contains code which is (more or less) cloned from
http://cvs.opensolaris.org/source/xref/usr/src/uts/common/os/swapgeneric.c#rootconf
The newboot rootconf() code in vfs.c contains a panic call which references the
string "svm_bootpath" at line 3889:
3885 error = VFS_MOUNTROOT(rootvfs, ROOT_INIT);
...
3889 if (error)
3890 cmn_err(CE_PANIC, "cannot mount root path %s",
svm_bootpath);
3891 return (error);
In my tests the "svm_bootpath" character array always contains an empty string,
so the panic message printed is "cannot mount root path ". I guess svm_bootpath
is a non-empty string only if svm (solaris volume manager ?) is used on the root
device.
The original code in swapgeneric.c seems to print more reasonble panic messages:
202 error = VFS_MOUNTROOT(rootvfs, ROOT_INIT);
...
206 if (error)
207 cmn_err(CE_CONT, "Cannot mount root on %s fstype %s\n",
208 rootfs.bo_name, rootfs.bo_fstype);
209 else
210 cmn_err(CE_CONT, "?root on %s fstype %s\n",
211 rootfs.bo_name, rootfs.bo_fstype);
212 return (error);
Why doesn't SPARC (swapgeneric.c rootconf()) report the svm_rootpath when
mounting
the root filesystem has failed? Or why must x86 newboot (vfs.c rootconf())
report
svm_rootpath when mounting the root fs has failed?
This message posted from opensolaris.org
