* [resend] 2.4: Support R4000 as a distinct CPU type
@ 2002-10-01 16:17 Maciej W. Rozycki
2002-10-02 13:47 ` Ralf Baechle
0 siblings, 1 reply; 3+ messages in thread
From: Maciej W. Rozycki @ 2002-10-01 16:17 UTC (permalink / raw)
To: Ralf Baechle; +Cc: linux-mips
Ralf,
Here is a new version that takes your recent mips64 cache code rearrange
into account. OK to apply?
BTW, how about renaming r5k-sc.c to sc-r5k.c for consistency?
Maciej
--
+ Maciej W. Rozycki, Technical University of Gdansk, Poland +
+--------------------------------------------------------------+
+ e-mail: macro@ds2.pg.gda.pl, PGP key available +
---------- Forwarded message ----------
Message-ID: <Pine.GSO.3.96.1020904172031.10619I-100000@delta.ds2.pg.gda.pl>
Date: Wed, 4 Sep 2002 17:57:15 +0200 (MET DST)
From: "Maciej W. Rozycki" <macro@ds2.pg.gda.pl>
To: Ralf Baechle <ralf@uni-koblenz.de>,
Karsten Merker <karsten@excalibur.cologne.de>,
linux-mips@linux-mips.org
Subject: [patch] 2.4.19: Support R4000 as a distinct CPU type
Hello,
Being an early implementation of a CPU family, R4000 contains a number of
interesting "features". One of them is a problem that affects a few
variations of shift operations (both 32-bit and 64-bit ones) that are
executed after multiply or divide instructions. Under certain
circumstances the shifts produce incorrect results. This is documented as
erratum #28 in "MIPS R4000PC/SC Errata, Processor Revision 2.2 and 3.0"
which is available at the MIPS site (I can't quote it here, sorry, as it's
copyrighted material and I failed to get permission from MIPS).
There is a partial workaround for the problem implemented in gcc which
gets activated if the "-mcpu=r4000" option is passed to the compiler.
Recent problems with the kernel when run on an affected CPU led me to
complete the multiplication part of the workaround (I'm also working on
the division part that is less biting). A suitable patch for 2.95.x is
available in mipsel-linux and mips64el-linux cross-gcc packages at my
site.
But to let gcc activate the workaround, kernel Makefiles must pass the
mentioned option. Here is a suitable patch. OK to apply?
I would like to express my gratitude to Karsten, who patiently tested on
his R4000 system a seemingly endless stream of kernels I fed him until I
discovered the reason of a mysterious misbehaviour. Without his help I
wouldn't be able to resolve the problem.
Maciej
--
+ Maciej W. Rozycki, Technical University of Gdansk, Poland +
+--------------------------------------------------------------+
+ e-mail: macro@ds2.pg.gda.pl, PGP key available +
patch-mips-2.4.20-pre6-20021001-r4000-2
diff -up --recursive --new-file linux-mips-2.4.20-pre6-20021001.macro/Documentation/Configure.help linux-mips-2.4.20-pre6-20021001/Documentation/Configure.help
--- linux-mips-2.4.20-pre6-20021001.macro/Documentation/Configure.help 2002-09-12 02:58:51.000000000 +0000
+++ linux-mips-2.4.20-pre6-20021001/Documentation/Configure.help 2002-09-30 23:28:19.000000000 +0000
@@ -2229,9 +2229,9 @@ CONFIG_PCMCIA_FDOMAIN
# Choice: mipstype
CPU type
-CONFIG_CPU_R3000
+CONFIG_CPU_MIPS32
Please make sure to pick the right CPU type. Linux/MIPS is not
- designed to be generic, i.e. Kernels compiled for R3000 CPUs will
+ designed to be generic, i.e. kernels compiled for R3000 CPUs will
*not* work on R4000 machines and vice versa. However, since most
of the supported machines have an R4000 (or similar) CPU, R4x00
might be a safe bet. If the resulting kernel does not work,
@@ -2243,10 +2243,12 @@ CONFIG_CPU_R3000
R6000 MIPS Technologies R6000-series processors,
including the 64474, 64475, 64574 and 64575.
+ R4000 MIPS Technologies R4000-series processors.
+
R4300 MIPS Technologies R4300-series processors.
- R4x00 MIPS Technologies R4000-series processors other than 4300,
- including the 4640, 4650, and 4700.
+ R4x00 MIPS Technologies R4xxx-series processors other than 4000
+ and 4300, including the 4640, 4650, and 4700.
R5000 MIPS Technologies R5000-series processors other than the
Nevada.
@@ -2260,14 +2262,18 @@ CONFIG_CPU_R6000
MIPS Technologies R6000-series processors, including the 64474,
64475, 64574 and 64575.
+R4000
+CONFIG_CPU_R4000
+ MIPS Technologies R4000-series processors.
+
R4300
CONFIG_CPU_R4300
MIPS Technologies R4300-series processors.
R4x00
CONFIG_CPU_R4X00
- MIPS Technologies R4000-series processors other than 4300, including
- the 4640, 4650, and 4700.
+ MIPS Technologies R4xxx-series processors other than 4000 and 4300,
+ including the 4640, 4650, and 4700.
R5000
CONFIG_CPU_R5000
diff -up --recursive --new-file linux-mips-2.4.20-pre6-20021001.macro/arch/mips/Makefile linux-mips-2.4.20-pre6-20021001/arch/mips/Makefile
--- linux-mips-2.4.20-pre6-20021001.macro/arch/mips/Makefile 2002-09-29 02:56:21.000000000 +0000
+++ linux-mips-2.4.20-pre6-20021001/arch/mips/Makefile 2002-09-30 23:28:19.000000000 +0000
@@ -57,6 +57,9 @@ endif
ifdef CONFIG_CPU_R6000
GCCFLAGS += -mcpu=r6000 -mips2 -Wa,--trap
endif
+ifdef CONFIG_CPU_R4000
+GCCFLAGS += -mcpu=r4000 -mips2 -Wa,--trap
+endif
ifdef CONFIG_CPU_R4300
GCCFLAGS += -mcpu=r4300 -mips2 -Wa,--trap
endif
diff -up --recursive --new-file linux-mips-2.4.20-pre6-20021001.macro/arch/mips/config-shared.in linux-mips-2.4.20-pre6-20021001/arch/mips/config-shared.in
--- linux-mips-2.4.20-pre6-20021001.macro/arch/mips/config-shared.in 2002-09-29 02:56:21.000000000 +0000
+++ linux-mips-2.4.20-pre6-20021001/arch/mips/config-shared.in 2002-09-30 23:28:19.000000000 +0000
@@ -445,6 +445,7 @@ choice 'CPU type' \
MIPS64 CONFIG_CPU_MIPS64 \
R3000 CONFIG_CPU_R3000 \
R39XX CONFIG_CPU_TX39XX \
+ R4000 CONFIG_CPU_R4000 \
R41xx CONFIG_CPU_VR41XX \
R4300 CONFIG_CPU_R4300 \
R4x00 CONFIG_CPU_R4X00 \
@@ -504,7 +505,8 @@ if [ "$CONFIG_CPU_SB1" = "y" ]; then
define_bool CONFIG_VTAG_ICACHE y
fi
-if [ "$CONFIG_CPU_R4X00" = "y" -o \
+if [ "$CONFIG_CPU_R4000" = "y" -o \
+ "$CONFIG_CPU_R4X00" = "y" -o \
"$CONFIG_CPU_R5000" = "y" -o \
"$CONFIG_CPU_RM7000" = "y" -o \
"$CONFIG_CPU_R10000" = "y" -o \
diff -up --recursive --new-file linux-mips-2.4.20-pre6-20021001.macro/arch/mips/dec/prom/init.c linux-mips-2.4.20-pre6-20021001/arch/mips/dec/prom/init.c
--- linux-mips-2.4.20-pre6-20021001.macro/arch/mips/dec/prom/init.c 2002-08-06 02:57:08.000000000 +0000
+++ linux-mips-2.4.20-pre6-20021001/arch/mips/dec/prom/init.c 2002-09-02 18:38:37.000000000 +0000
@@ -100,7 +100,7 @@ int __init prom_init(int argc, char **ar
}
#endif
-#if defined(CONFIG_CPU_R4X00)
+#if defined(CONFIG_CPU_R4000) || defined(CONFIG_CPU_R4X00)
if ((mips_cpu.cputype == CPU_R3000) ||
(mips_cpu.cputype == CPU_R3000A)) {
prom_printf("Sorry, this kernel is compiled for the wrong CPU type!\n");
diff -up --recursive --new-file linux-mips-2.4.20-pre6-20021001.macro/arch/mips/mm/Makefile linux-mips-2.4.20-pre6-20021001/arch/mips/mm/Makefile
--- linux-mips-2.4.20-pre6-20021001.macro/arch/mips/mm/Makefile 2002-09-03 02:56:40.000000000 +0000
+++ linux-mips-2.4.20-pre6-20021001/arch/mips/mm/Makefile 2002-09-30 23:28:19.000000000 +0000
@@ -17,6 +17,7 @@ obj-$(CONFIG_CPU_R3000) += pg-r3k.o c-r
tlbex-r3k.o
obj-$(CONFIG_CPU_TX39XX) += pg-r3k.o c-tx39.o tlb-r3k.o tlbex-r3k.o
obj-$(CONFIG_CPU_TX49XX) += pg-r4k.o c-tx49.o tlb-r4k.o tlbex-r4k.o
+obj-$(CONFIG_CPU_R4000) += pg-r4k.o c-r4k.o tlb-r4k.o tlbex-r4k.o
obj-$(CONFIG_CPU_R4300) += pg-r4k.o c-r4k.o tlb-r4k.o tlbex-r4k.o
obj-$(CONFIG_CPU_R4X00) += pg-r4k.o c-r4k.o tlb-r4k.o tlbex-r4k.o
obj-$(CONFIG_CPU_VR41XX) += pg-r4k.o c-r4k.o tlb-r4k.o tlbex-r4k.o
diff -up --recursive --new-file linux-mips-2.4.20-pre6-20021001.macro/arch/mips/mm/loadmmu.c linux-mips-2.4.20-pre6-20021001/arch/mips/mm/loadmmu.c
--- linux-mips-2.4.20-pre6-20021001.macro/arch/mips/mm/loadmmu.c 2002-08-06 02:57:32.000000000 +0000
+++ linux-mips-2.4.20-pre6-20021001/arch/mips/mm/loadmmu.c 2002-09-02 18:42:30.000000000 +0000
@@ -68,9 +68,9 @@ extern void sb1_tlb_init(void);
void __init loadmmu(void)
{
if (mips_cpu.options & MIPS_CPU_4KTLB) {
-#if defined(CONFIG_CPU_R4X00) || defined(CONFIG_CPU_VR41XX) || \
- defined(CONFIG_CPU_R4300) || defined(CONFIG_CPU_R5000) || \
- defined(CONFIG_CPU_NEVADA)
+#if defined(CONFIG_CPU_R4000) || defined(CONFIG_CPU_VR41XX) || \
+ defined(CONFIG_CPU_R4300) || defined(CONFIG_CPU_R4X00) || \
+ defined(CONFIG_CPU_R5000) || defined(CONFIG_CPU_NEVADA)
ld_mmu_r4xx0();
r4k_tlb_init();
#endif
diff -up --recursive --new-file linux-mips-2.4.20-pre6-20021001.macro/arch/mips64/Makefile linux-mips-2.4.20-pre6-20021001/arch/mips64/Makefile
--- linux-mips-2.4.20-pre6-20021001.macro/arch/mips64/Makefile 2002-09-03 02:58:10.000000000 +0000
+++ linux-mips-2.4.20-pre6-20021001/arch/mips64/Makefile 2002-09-30 23:28:19.000000000 +0000
@@ -46,6 +46,9 @@ endif
#
# CPU-dependent compiler/assembler options for optimization.
#
+ifdef CONFIG_CPU_R4000
+GCCFLAGS += -mcpu=r4000 -mips3
+endif
ifdef CONFIG_CPU_R4300
GCCFLAGS += -mcpu=r4300 -mips3
endif
diff -up --recursive --new-file linux-mips-2.4.20-pre6-20021001.macro/arch/mips64/mm/Makefile linux-mips-2.4.20-pre6-20021001/arch/mips64/mm/Makefile
--- linux-mips-2.4.20-pre6-20021001.macro/arch/mips64/mm/Makefile 2002-10-01 02:56:55.000000000 +0000
+++ linux-mips-2.4.20-pre6-20021001/arch/mips64/mm/Makefile 2002-10-01 16:11:06.000000000 +0000
@@ -10,6 +10,7 @@ export-objs += umap.o
obj-y := extable.o init.o fault.o loadmmu.o \
tlb-glue-r4k.o tlbex-r4k.o
+obj-$(CONFIG_CPU_R4000) += c-r4k.o pg-r4k.o tlb-r4k.o
obj-$(CONFIG_CPU_R4300) += c-r4k.o pg-r4k.o tlb-r4k.o
obj-$(CONFIG_CPU_R4X00) += c-r4k.o pg-r4k.o tlb-r4k.o
obj-$(CONFIG_CPU_R5000) += c-r4k.o pg-r4k.o tlb-r4k.o r5k-sc.o
diff -up --recursive --new-file linux-mips-2.4.20-pre6-20021001.macro/arch/mips64/mm/loadmmu.c linux-mips-2.4.20-pre6-20021001/arch/mips64/mm/loadmmu.c
--- linux-mips-2.4.20-pre6-20021001.macro/arch/mips64/mm/loadmmu.c 2002-09-29 02:56:35.000000000 +0000
+++ linux-mips-2.4.20-pre6-20021001/arch/mips64/mm/loadmmu.c 2002-09-30 23:28:19.000000000 +0000
@@ -60,7 +60,8 @@ extern void r4k_tlb_init(void);
void __init load_mmu(void)
{
if (mips_cpu.options & MIPS_CPU_4KTLB) {
-#if defined (CONFIG_CPU_R4300) \
+#if defined (CONFIG_CPU_R4000) \
+ || defined (CONFIG_CPU_R4300) \
|| defined (CONFIG_CPU_R4X00) \
|| defined (CONFIG_CPU_R5000) \
|| defined (CONFIG_CPU_NEVADA)
diff -up --recursive --new-file linux-mips-2.4.20-pre6-20021001.macro/include/asm-mips64/addrspace.h linux-mips-2.4.20-pre6-20021001/include/asm-mips64/addrspace.h
--- linux-mips-2.4.20-pre6-20021001.macro/include/asm-mips64/addrspace.h 2002-07-26 02:58:05.000000000 +0000
+++ linux-mips-2.4.20-pre6-20021001/include/asm-mips64/addrspace.h 2002-10-01 16:10:29.000000000 +0000
@@ -82,7 +82,8 @@
#define CKSSEG 0xffffffffc0000000
#define CKSEG3 0xffffffffe0000000
-#if defined (CONFIG_CPU_R4300) \
+#if defined (CONFIG_CPU_R4000) \
+ || defined (CONFIG_CPU_R4300) \
|| defined (CONFIG_CPU_R4X00) \
|| defined (CONFIG_CPU_R5000) \
|| defined (CONFIG_CPU_NEVADA) \
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [resend] 2.4: Support R4000 as a distinct CPU type
2002-10-01 16:17 [resend] 2.4: Support R4000 as a distinct CPU type Maciej W. Rozycki
@ 2002-10-02 13:47 ` Ralf Baechle
2002-10-02 14:18 ` Maciej W. Rozycki
0 siblings, 1 reply; 3+ messages in thread
From: Ralf Baechle @ 2002-10-02 13:47 UTC (permalink / raw)
To: Maciej W. Rozycki; +Cc: linux-mips
On Tue, Oct 01, 2002 at 06:17:11PM +0200, Maciej W. Rozycki wrote:
> Here is a new version that takes your recent mips64 cache code rearrange
> into account. OK to apply?
I'm not sure if that's really a good idea. Technically it's ok but I expect
users of the R4000 to missconfigure their kernels. So I wonder if it might
be more appropriate to have just automatically enabled this workaround for
systems that are affected? If we keep it user-selectable then we at least
want a safety check somewhere in the startup code telling users to rebuild
their code with the workaround enabled.
Having this workaround enabled by default would also ensure Linux
distributions ship working code - you don't want users having to recompile
their whole distribution ...
> BTW, how about renaming r5k-sc.c to sc-r5k.c for consistency?
Yes. Though there's a bit of confusion hidden there - the Indy SC code
is named ip22-sc.c - but that at least in a different directory.
Ralf
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [resend] 2.4: Support R4000 as a distinct CPU type
2002-10-02 13:47 ` Ralf Baechle
@ 2002-10-02 14:18 ` Maciej W. Rozycki
0 siblings, 0 replies; 3+ messages in thread
From: Maciej W. Rozycki @ 2002-10-02 14:18 UTC (permalink / raw)
To: Ralf Baechle; +Cc: linux-mips
On Wed, 2 Oct 2002, Ralf Baechle wrote:
> > Here is a new version that takes your recent mips64 cache code rearrange
> > into account. OK to apply?
>
> I'm not sure if that's really a good idea. Technically it's ok but I expect
> users of the R4000 to missconfigure their kernels. So I wonder if it might
I plan to write some code to detect various processor bugs, including
this one. A nice kernel panic will instruct users how to configure the
kernel properly, so this is not an issue. If you want me to defer the
patch until it's done, I see no problem. Others may apply the patch as
needed for now.
> be more appropriate to have just automatically enabled this workaround for
> systems that are affected? If we keep it user-selectable then we at least
The workaround affects gcc's code generation -- there is no way to change
it at the run time.
> want a safety check somewhere in the startup code telling users to rebuild
> their code with the workaround enabled.
Of course, as I've written above.
> Having this workaround enabled by default would also ensure Linux
> distributions ship working code - you don't want users having to recompile
> their whole distribution ...
But the generated code is worse -- it interlocks on a multiplication
flushing the whole benefit of the separate multiply unit down the drain.
There is room for an improvement here, though.
As to compiling userland -- the workaround is the gcc's default for R4000
which is what is selected by "-mips3" among others. It should probably be
enabled by default for "-mips1" and "-mips2" configurations as well if no
explicit CPU selection options are passed.
Ultimately we'll want to have a separate setting for the R4400 in the
kernel as well, due to a smaller set of bugs.
Maciej
--
+ Maciej W. Rozycki, Technical University of Gdansk, Poland +
+--------------------------------------------------------------+
+ e-mail: macro@ds2.pg.gda.pl, PGP key available +
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2002-10-02 14:17 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2002-10-01 16:17 [resend] 2.4: Support R4000 as a distinct CPU type Maciej W. Rozycki
2002-10-02 13:47 ` Ralf Baechle
2002-10-02 14:18 ` Maciej W. Rozycki
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox