On 11.09.2011 15:14, peter fuerst wrote:
Here's an attempt to bloat the linux source a bit more ;-)
This patch brings, yet missing, parts that make a Linux-driven Indigo2
Impact (IP28 and most probably IP22-Impact) an usable desktop-machine
"out of the box".
The driver provides the framebuffer console and an interface for the
Xserver (mmap'ing a DMA-pool to the shadow framebuffer and doing the
necessary cacheflush).
Meanwhile only a few files are affected and obviously no side-effects
to other parts of the kernel are to be expected.
BTW: it would be appreciated, if someone could verify, that this driver
also works for IP22 Impact.
Signed-off-by: peter fuerst <post@xxxxxxxx>
---
1) Preparation
drivers/video/Kconfig | 6 ++++++
drivers/video/Makefile | 1 +
drivers/video/logo/Kconfig | 2 +-
arch/mips/sgi-ip22/ip22-setup.c | 37 +++++++++++++++++++++++++++++++++++++
arch/mips/configs/ip28_defconfig | 1 +
5 files changed, 46 insertions(+), 1 deletions(-)
diff --git a/drivers/video/Kconfig b/drivers/video/Kconfig
index 549b960..43dbc9f 100644
--- a/drivers/video/Kconfig Thu Jan 1 00:00:00 1970
+++ b/drivers/video/Kconfig Thu Mar 10 23:37:58 2011
@@ -2383,6 +2383,12 @@ config FB_PUV3_UNIGFX
Choose this option if you want to use the Unigfx device as a
framebuffer device. Without the support of PCI & AGP.
It looks like the patch is spoiled as I'm seeing two spaces at the start
of line when looking at the message source.
+config FB_IMPACT
+ tristate "SGI Indigo2 Impact graphics support"
+ depends on FB && (SGI_IP22 || SGI_IP28 || SGI_IP30)
+ help
+ SGI Indigo2 Impact (SI/HI/MI) graphics card support.
+
source "drivers/video/omap/Kconfig"
source "drivers/video/omap2/Kconfig"
There are alos empty lines after each file in the patch -- which shouldn't
be there.
diff --git a/drivers/video/Makefile b/drivers/video/Makefile
index 8b83129..278c8fa 100644
--- a/drivers/video/Makefile Thu Jan 1 00:00:00 1970
+++ b/drivers/video/Makefile Thu Mar 10 23:44:40 2011
@@ -141,6 +141,7 @@ obj-$(CONFIG_FB_MSM) += msm/
obj-$(CONFIG_FB_NUC900) += nuc900fb.o
obj-$(CONFIG_FB_JZ4740) += jz4740_fb.o
obj-$(CONFIG_FB_PUV3_UNIGFX) += fb-puv3.o
+obj-$(CONFIG_FB_IMPACT) += impact.o
# Platform or fallback drivers go here
obj-$(CONFIG_FB_UVESA) += uvesafb.o
The above should be a part of the driver patch, as you can't add Makefile
targets fow which no source files exist yet.
diff --git a/arch/mips/sgi-ip22/ip22-setup.c b/arch/mips/sgi-ip22/ip22-setup.c
index 5e66213..085b612 100644
--- a/arch/mips/sgi-ip22/ip22-setup.c Thu Jan 1 00:00:00 1970
+++ b/arch/mips/sgi-ip22/ip22-setup.c Fri May 8 01:05:13 2009
@@ -96,4 +96,41 @@ void __init plat_mem_setup(void)
}
}
#endif
+#if defined(CONFIG_FB_IMPACT)
+ {
+ extern void setup_impact_earlycons(void);
+ /*
+ * Get graphics info before it is overwritten...
+ * E.g. @ 9000000020f02f78: ffffffff9fc6d770,900000001f000000
+ */
+#ifdef CONFIG_ARC64
+ ULONG * (*__vec)(void) = (typeof(__vec))
+ ((ULONG*)PROMBLOCK->pvector)[8];
+ ULONG *gfxinfo = (*__vec)();
+#else
+ /* supposed to work on both 32/64-bit kernels. */
+ int (*__vec)(void) = (typeof(__vec))
+ (long) ((int*)PROMBLOCK->pvector)[8];
+ int *gfxinfo = (typeof(gfxinfo)) (*__vec)();
+#endif
+ /* See note on __pa() in impact.c */
+ sgi_gfxaddr = __pa((void*)gfxinfo[1]);
+ if (sgi_gfxaddr < 0x1f000000 || 0x1fa00000 <= sgi_gfxaddr)
Immediate should generally be the right operand in comparison.
2) The driver
drivers/video/impact.c | 1154 ++++++++++++++++++++++++++++++++++++++++++++++++
include/video/impact.h | 210 +++++++++
2 files changed, 1364 insertions(+), 0 deletions(-)
You should have posted it in a separate mail.
diff --git a/drivers/video/impact.c b/drivers/video/impact.c
new file mode 100644
index 0000000..5ffc948
--- /dev/null Wed Dec 8 00:46:04 2004
+++ b/drivers/video/impact.c Mon Jul 25 00:06:58 2011
@@ -0,0 +1,1154 @@
I lack the time to review such a large driver, unfortunately...
WBR, Sergei
