I'm trying to debug a reported problem with filemon(4), and to make
things a bit easier I tried to pass STDOUT_FILENO as the place where
filemon should write its output. Running this on a "console terminal"
within qemu, the code in filemon_output() sets up a struct iovec and
initializes a struct uio by calling uio_setup_sysspace(iovec). It then
calls (*filemon->fm_fp->f_ops->fo_write) which ends up in spec_write().
For some reason, spec_write() doesn't like something, and the code at
#ifdef DIAGNOSTIC
if (uio->uio_rw != UIO_WRITE)
panic("spec_write mode");
if (&uio->uio_vmspace->vm_map != kernel_map &&
uio->uio_vmspace != curproc->p_vmspace)
panic("spec_write proc");
#endif
ends up by calling the panic("spec_write proc").
It's not clear to me what the DIAGNOSTIC code is supposed to prevent,
nor is it clear why the call to uio_setup_sysspace() doesn't set up
everything needed.
The same code in filemon seems to work just fine when it is passed a
file descriptor that refers to a "real" file.
Can someone enlighten me here?
+------------------+--------------------------+------------------------+
| Paul Goyette | PGP Key fingerprint: | E-mail addresses: |
| (Retired) | FA29 0E3B 35AF E8AE 6651 | paul at whooppee.com |
| Kernel Developer | 0786 F758 55DE 53BA 7731 | pgoyette at netbsd.org |
+------------------+--------------------------+------------------------+
