[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Qemu-devel] [PATCH] take2 SH: On-chip PCI controller support.
From: |
takasi-y |
Subject: |
[Qemu-devel] [PATCH] take2 SH: On-chip PCI controller support. |
Date: |
Wed, 12 Nov 2008 07:58:29 +0900 (JST) |
Thank you for your review.
> > +++ b/hw/sh_pci.c
...
> > + switch(addr) {
> > + case 0 ... 0xff: *(uint32_t*)(pcic->dev->config + addr) = val;
>
> Is a break; missing above?
Oops, mistake. Perhaps, this excess compact format is bad. Fixed.
> You should convert endianness here and in sh_pci_reg_read below.
I fixed it with le32_to_cpu/cpu_to_le32.
What needed is "cpu(big or little) <-> pci(little)", isn't it?
> Also, just wondering, are unaligned accesses possible? If this
> function is called with addr == 0xff,
> 1. an unaligned access on the host happens,
> 2. pcic->dev->config[259] is accessed, and there are only 256 elements.
I don't think unaligned accesses will happen here on sh4, (emulated)CPU will
catch unaligned access before here. But anyway, 0xfc seems to be better.
> > @@ -432,6 +432,9 @@ int get_physical_address(CPUState * env, target_ulong *
> > physical,
> > if (address >= 0x80000000 && address < 0xc0000000) {
> > /* Mask upper 3 bits for P1 and P2 areas */
> > *physical = address & 0x1fffffff;
> > + } else if (address >= 0xfd000000 && address < 0xfe000000) {
> > + /* PCI memory space */
> > + *physical = address;
>
> Hmm, this doesn't look right, but I'm clueless about the sh4
> architecture... If top 3 bits are masked, will it be possible to
> differentiate between P1, P2, etc areas?
The difference between P1 and P2 is cache on/off.
Because we don't have cache implemented, it is OK, __so far__.
> ... It seems none of them is \n currently implemented (?),
Perhaps you are right. I was surprised when I saw this code first.
But after trying to "fix" it. I found it is a reasonable solution.
> ... but if they are, I think they should simply
> register the area multiple times at different offsets to make it as if
> the top three bits had been masked.
I think so, too.
Another mapping P4(0xff*) to A7(0x1f*) just below there
| if (address >= 0xfc000000)
| *physical = address & 0x1fffffff;
is something wrong, but works well, too.
# A7 is kind of magic address on TLB which is translated to P4.
After I
- register 3 times(P2,P1,Physical) for each ROM,RAM,VRAM, and LSIs.
- register 2 times(P4,A7) for each internal modules.
- modify codes of sh modules to work on both address P4, and A7.
I successfully boot linux without these translation.
But, I found I have to do something with all modules which initialize
routine does "init + alloc + register" (to add extra registration).
Possibly, I can do it with modified cpu_register_physical_memory().
But, it seems to be common routine for all architectures.
Are there any case using special registration routine?
> ... This is not related to the PCI \n implementation ofcourse.
Yes. I'd like this patch to be merged separately of this address
translation issue. I think I need more time to solve it.
/yoshii
---
This patch adds SuperH on-chip PCI controller(PCIC) support.
Signed-off-by: Takashi YOSHII <address@hidden>
---
Makefile.target | 2 +-
hw/sh_pci.c | 207 +++++++++++++++++++++++++++++++++++++++++++++++++++
target-sh4/helper.c | 3 +
3 files changed, 211 insertions(+), 1 deletions(-)
create mode 100644 hw/sh_pci.c
diff --git a/Makefile.target b/Makefile.target
index fdaaa67..0f4af35 100644
--- a/Makefile.target
+++ b/Makefile.target
@@ -737,7 +737,7 @@ CPPFLAGS += -DHAS_AUDIO
endif
ifeq ($(TARGET_BASE_ARCH), sh4)
OBJS+= shix.o r2d.o sh7750.o sh7750_regnames.o tc58128.o
-OBJS+= sh_timer.o ptimer.o sh_serial.o sh_intc.o sm501.o serial.o
+OBJS+= sh_timer.o ptimer.o sh_serial.o sh_intc.o sh_pci.o sm501.o serial.o
OBJS+= ide.o
endif
ifeq ($(TARGET_BASE_ARCH), m68k)
diff --git a/hw/sh_pci.c b/hw/sh_pci.c
new file mode 100644
index 0000000..c063fa4
--- /dev/null
+++ b/hw/sh_pci.c
@@ -0,0 +1,207 @@
+/*
+ * SuperH on-chip PCIC emulation.
+ *
+ * Copyright (c) 2008 Takashi YOSHII
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+#include "hw.h"
+#include "sh.h"
+#include "pci.h"
+#include "bswap.h"
+
+typedef struct {
+ PCIBus *bus;
+ PCIDevice *dev;
+ uint32_t regbase;
+ uint32_t iopbase;
+ uint32_t membase;
+ uint32_t par;
+ uint32_t mbr;
+ uint32_t iobr;
+} SHPCIC;
+
+static void sh_pci_reg_write (void *p, target_phys_addr_t addr, uint32_t val)
+{
+ SHPCIC *pcic = p;
+ addr -= pcic->regbase;
+ switch(addr) {
+ case 0 ... 0xfc:
+ cpu_to_le32w((uint32_t*)(pcic->dev->config + addr), val);
+ break;
+ case 0x1c0:
+ pcic->par = val;
+ break;
+ case 0x1c4:
+ pcic->mbr = val;
+ break;
+ case 0x1c8:
+ pcic->iobr = val;
+ break;
+ case 0x220:
+ pci_data_write(pcic->bus, pcic->par, val, 4);
+ break;
+ }
+}
+
+static uint32_t sh_pci_reg_read (void *p, target_phys_addr_t addr)
+{
+ SHPCIC *pcic = p;
+ addr -= pcic->regbase;
+ switch(addr) {
+ case 0 ... 0xfc:
+ return le32_to_cpup((uint32_t*)(pcic->dev->config + addr));
+ case 0x1c0:
+ return pcic->par;
+ case 0x220:
+ return pci_data_read(pcic->bus, pcic->par, 4);
+ }
+ return 0;
+}
+
+static void sh_pci_data_write (SHPCIC *pcic, target_phys_addr_t addr,
+ uint32_t val, int size)
+{
+ pci_data_write(pcic->bus, addr - pcic->membase + pcic->mbr, val, size);
+}
+
+static uint32_t sh_pci_mem_read (SHPCIC *pcic, target_phys_addr_t addr,
+ int size)
+{
+ return pci_data_read(pcic->bus, addr - pcic->membase + pcic->mbr, size);
+}
+
+static void sh_pci_writeb (void *p, target_phys_addr_t addr, uint32_t val)
+{
+ sh_pci_data_write(p, addr, val, 1);
+}
+
+static void sh_pci_writew (void *p, target_phys_addr_t addr, uint32_t val)
+{
+ sh_pci_data_write(p, addr, val, 2);
+}
+
+static void sh_pci_writel (void *p, target_phys_addr_t addr, uint32_t val)
+{
+ sh_pci_data_write(p, addr, val, 4);
+}
+
+static uint32_t sh_pci_readb (void *p, target_phys_addr_t addr)
+{
+ return sh_pci_mem_read(p, addr, 1);
+}
+
+static uint32_t sh_pci_readw (void *p, target_phys_addr_t addr)
+{
+ return sh_pci_mem_read(p, addr, 2);
+}
+
+static uint32_t sh_pci_readl (void *p, target_phys_addr_t addr)
+{
+ return sh_pci_mem_read(p, addr, 4);
+}
+
+static int sh_pci_addr2port(SHPCIC *pcic, target_phys_addr_t addr)
+{
+ return addr - pcic->iopbase + pcic->iobr;
+}
+
+static void sh_pci_outb (void *p, target_phys_addr_t addr, uint32_t val)
+{
+ cpu_outb(NULL, sh_pci_addr2port(p, addr), val);
+}
+
+static void sh_pci_outw (void *p, target_phys_addr_t addr, uint32_t val)
+{
+ cpu_outw(NULL, sh_pci_addr2port(p, addr), val);
+}
+
+static void sh_pci_outl (void *p, target_phys_addr_t addr, uint32_t val)
+{
+ cpu_outl(NULL, sh_pci_addr2port(p, addr), val);
+}
+
+static uint32_t sh_pci_inb (void *p, target_phys_addr_t addr)
+{
+ return cpu_inb(NULL, sh_pci_addr2port(p, addr));
+}
+
+static uint32_t sh_pci_inw (void *p, target_phys_addr_t addr)
+{
+ return cpu_inw(NULL, sh_pci_addr2port(p, addr));
+}
+
+static uint32_t sh_pci_inl (void *p, target_phys_addr_t addr)
+{
+ return cpu_inl(NULL, sh_pci_addr2port(p, addr));
+}
+
+typedef struct {
+ CPUReadMemoryFunc *r[3];
+ CPUWriteMemoryFunc *w[3];
+} MemOp;
+
+static MemOp sh_pci_reg = {
+ { NULL, NULL, sh_pci_reg_read },
+ { NULL, NULL, sh_pci_reg_write },
+};
+
+static MemOp sh_pci_mem = {
+ { sh_pci_readb, sh_pci_readw, sh_pci_readl },
+ { sh_pci_writeb, sh_pci_writew, sh_pci_writel },
+};
+
+static MemOp sh_pci_iop = {
+ { sh_pci_inb, sh_pci_inw, sh_pci_inl },
+ { sh_pci_outb, sh_pci_outw, sh_pci_outl },
+};
+
+PCIBus *sh_pci_register_bus(pci_set_irq_fn set_irq, pci_map_irq_fn map_irq,
+ qemu_irq *pic, int devfn_min, int nirq)
+{
+ SHPCIC *p;
+ int mem, reg, iop;
+
+ p = qemu_mallocz(sizeof(SHPCIC));
+ p->bus = pci_register_bus(set_irq, map_irq, pic, devfn_min, nirq);
+
+ p->dev = pci_register_device(p->bus, "SH PCIC", sizeof(PCIDevice),
+ -1, NULL, NULL);
+ p->regbase = 0x1e200000;
+ p->iopbase = 0x1e240000;
+ p->membase = 0xfd000000;
+ reg = cpu_register_io_memory(0, sh_pci_reg.r, sh_pci_reg.w, p);
+ mem = cpu_register_io_memory(0, sh_pci_mem.r, sh_pci_mem.w, p);
+ iop = cpu_register_io_memory(0, sh_pci_iop.r, sh_pci_iop.w, p);
+ cpu_register_physical_memory(p->regbase, 0x224, reg);
+ cpu_register_physical_memory(p->iopbase, 0x40000, iop);
+ cpu_register_physical_memory(p->membase, 0x1000000, mem);
+
+ p->dev->config[0x00] = 0x54; // HITACHI
+ p->dev->config[0x01] = 0x10; //
+ p->dev->config[0x02] = 0x0e; // SH7751R
+ p->dev->config[0x03] = 0x35; //
+ p->dev->config[0x04] = 0x80;
+ p->dev->config[0x05] = 0x00;
+ p->dev->config[0x06] = 0x90;
+ p->dev->config[0x07] = 0x02;
+
+ return p->bus;
+}
+
diff --git a/target-sh4/helper.c b/target-sh4/helper.c
index 6429862..a63bde7 100644
--- a/target-sh4/helper.c
+++ b/target-sh4/helper.c
@@ -432,6 +432,9 @@ int get_physical_address(CPUState * env, target_ulong *
physical,
if (address >= 0x80000000 && address < 0xc0000000) {
/* Mask upper 3 bits for P1 and P2 areas */
*physical = address & 0x1fffffff;
+ } else if (address >= 0xfd000000 && address < 0xfe000000) {
+ /* PCI memory space */
+ *physical = address;
} else if (address >= 0xfc000000) {
/*
* Mask upper 3 bits for control registers in P4 area,
--
1.5.6.3