[Qemu-devel] [PATCH] Fixing sh4 serial abort

[Rob Landley <rob@landley.net>]

If you grab the current aboriginal linux build scripts:

  http://landley.net/hg/aboriginal/archive/tip.tar.bz2

And "./build.sh sh4", then cd to build/system-image-sh4 and
"./run-emulator.sh" you get this:

  sh_serial: unsupported read from 0x18
  Aborted

The bug was triggered by linux kernel commit 73c3d53f38e0a8e6 back
between v3.2 and v3.3, which did this:

--- a/drivers/tty/serial/sh-sci.c
+++ b/drivers/tty/serial/sh-sci.c
@@ -1771,18 +1771,25 @@ static void sci_set_termios(struct uart_port
*port, struct ktermios *termios,

 	sci_init_pins(port, termios->c_cflag);

-	if (s->cfg->capabilities & SCIx_HAVE_RTSCTS) {
-		reg = sci_getreg(port, SCFCR);
-		if (reg->size) {
-			unsigned short ctrl;
+	reg = sci_getreg(port, SCFCR);
+	if (reg->size) {
+		unsigned short ctrl = sci_in(port, SCFCR);

-			ctrl = sci_in(port, SCFCR);
+		if (s->cfg->capabilities & SCIx_HAVE_RTSCTS) {

I.E. sci_getreg(port, SCFCR) move to before checking whether or not
we'll ever possibly use the result. SCFCR is 0x18 and QEMU calls abort()
on an attempt to read from an unimplemented register.

I can patch the kernel to work around this (and probably will for this
release), but the _proper_ fix is to get qemu not to abort on a register
read that works fine if it just returns 0.

It turns out the qemu fix (in current git) is just:

--- a/hw/sh_serial.c
+++ b/hw/sh_serial.c
@@ -248,11 +248,9 @@ static uint64_t sh_serial_read(void *opaque,
target_phys
                     s->flags &= ~SH_SERIAL_FLAG_RDF;
             }
             break;
-#if 0
         case 0x18:
             ret = s->fcr;
             break;
-#endif
         case 0x1c:
             ret = s->rx_cnt;
             break;

(It can also just do ret = 0; and that works too. Or comment out the
abort() near the end of the function.)

Doing that, qemu boots the sh4 system image to a shell prompt, and the
result compiles "hello world" natively.

By the way, this board emulation (r2d) only has 64 megs ram. Is there an
easy way to get 256 megs out of it?

Rob
-- 
GNU/Linux isn't: Linux=GPLv2, GNU=GPLv3+, they can't share code.
Either it's "mere aggregation", or a license violation.  Pick one.