Skip to contentSkip to conversation

m68k compile issue with 4.0.5

DISCUSSION Waldemar Brodkorb 3 participants 7 messages Jun 15, 2015 – Jun 18, 2015

lore
Share & tools
lore
#0 lore raw

Hi,

I am trying to build a M68K (Coldfire no-MMU) kernel for Qemu-system-m68k.

With 4.0.4 everything is fine. With 4.0.5 I get following compile
error:
adk-uclinux-gcc -Wp,-MD,mm/.nommu.o.d -nostdinc -isystem
/home/wbx/m68k/toolchain_qemu-m68k_uclibc-ng_m68k_nommu/usr/lib/gcc/m68k-openadk-uclinux-uclibc/4.9.2/include -I./arch/m68k/include -Iarch/m68k/include/generated/uapi
-Iarch/m68k/include/generated -Iinclude -I./arch/m68k/include/uapi -Iarch/m68k/include/generated/uapi -I./include/uapi
-Iinclude/generated/uapi -include ./include/linux/kconfig.h
-D__KERNEL__ -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs
-fno-strict-aliasing -fno-common
-Werror-implicit-function-declaration -Wno-format-security
-std=gnu89 -mcpu=5208 -pipe -DUTS_SYSNAME=\"uClinux\" -D__uClinux__ -fno-delete-null-pointer-checks -Os -Wno-maybe-uninitialized --param=allow-store-data-races=0 -Wframe-larger-than=1024
-fno-stack-protector -Wno-unused-but-set-variable
-fomit-frame-pointer -fno-var-tracking-assignments
-Wdeclaration-after-statement -Wno-pointer-sign -fno-strict-overflow
-fconserve-stack -Werror=implicit-int -Werror=strict-prototypes
-Werror=date-time -DCC_HAVE_ASM_GOTO -D"KBUILD_STR(s)=#s"
-D"KBUILD_BASENAME=KBUILD_STR(nommu)"
-D"KBUILD_MODNAME=KBUILD_STR(nommu)" -c -o mm/nommu.o mm/nommu.c
mm/nommu.c: In function 'delete_vma':
mm/nommu.c:861:3: error: implicit declaration of function 'vma_fput' [-Werror=implicit-function-declaration]
vma_fput(vma);
^
cc1: some warnings being treated as errors

Any idea what change breaks the compile?

best regards
Waldemar

#1 lore raw

Hi Waldemar,

Quoted 26 lines · Waldemar Brodkorb

I am trying to build a M68K (Coldfire no-MMU) kernel for Qemu-system-m68k.

With 4.0.4 everything is fine. With 4.0.5 I get following compile
error:
adk-uclinux-gcc -Wp,-MD,mm/.nommu.o.d -nostdinc -isystem
/home/wbx/m68k/toolchain_qemu-m68k_uclibc-ng_m68k_nommu/usr/lib/gcc/m68k-openadk-uclinux-uclibc/4.9.2/include

(quoted diff) +0 -17
-I./arch/m68k/include -Iarch/m68k/include/generated/uapi
-Iarch/m68k/include/generated  -Iinclude -I./arch/m68k/include/uapi
-Iarch/m68k/include/generated/uapi -I./include/uapi
-Iinclude/generated/uapi -include ./include/linux/kconfig.h
-D__KERNEL__ -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs
-fno-strict-aliasing -fno-common
-Werror-implicit-function-declaration -Wno-format-security
-std=gnu89 -mcpu=5208 -pipe -DUTS_SYSNAME=\"uClinux\" -D__uClinux__
-fno-delete-null-pointer-checks -Os -Wno-maybe-uninitialized
--param=allow-store-data-races=0 -Wframe-larger-than=1024
-fno-stack-protector -Wno-unused-but-set-variable
-fomit-frame-pointer -fno-var-tracking-assignments
-Wdeclaration-after-statement -Wno-pointer-sign -fno-strict-overflow
-fconserve-stack -Werror=implicit-int -Werror=strict-prototypes
-Werror=date-time -DCC_HAVE_ASM_GOTO    -D"KBUILD_STR(s)=#s"
-D"KBUILD_BASENAME=KBUILD_STR(nommu)"
-D"KBUILD_MODNAME=KBUILD_STR(nommu)" -c -o mm/nommu.o mm/nommu.c

mm/nommu.c: In function 'delete_vma':
mm/nommu.c:861:3: error: implicit declaration of function 'vma_fput' [-Werror=implicit-function-declaration]

Quoted 5 lines
   vma_fput(vma);
   ^

cc1: some warnings being treated as errors

Any idea what change breaks the compile?

I tried a few m68knommu defconfigs, but can't reproduce it.

Is this a plain v4.0.5? I can't find the offending call to vma_fput(). "git grep vma_fput" tells me there's no "vma_fput" in the kernel sources?

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#2 lore raw

Hi Waldemar,

Quoted 33 lines · Geert Uytterhoeven

Hi Waldemar,

Quoted 30 lines · Waldemar Brodkorb

I am trying to build a M68K (Coldfire no-MMU) kernel for Qemu-system-m68k.

With 4.0.4 everything is fine. With 4.0.5 I get following compile
error:
adk-uclinux-gcc -Wp,-MD,mm/.nommu.o.d -nostdinc -isystem
/home/wbx/m68k/toolchain_qemu-m68k_uclibc-ng_m68k_nommu/usr/lib/gcc/m68k-openadk-uclinux-uclibc/4.9.2/include

(quoted diff) +0 -17
-I./arch/m68k/include -Iarch/m68k/include/generated/uapi
-Iarch/m68k/include/generated  -Iinclude -I./arch/m68k/include/uapi
-Iarch/m68k/include/generated/uapi -I./include/uapi
-Iinclude/generated/uapi -include ./include/linux/kconfig.h
-D__KERNEL__ -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs
-fno-strict-aliasing -fno-common
-Werror-implicit-function-declaration -Wno-format-security
-std=gnu89 -mcpu=5208 -pipe -DUTS_SYSNAME=\"uClinux\" -D__uClinux__
-fno-delete-null-pointer-checks -Os -Wno-maybe-uninitialized
--param=allow-store-data-races=0 -Wframe-larger-than=1024
-fno-stack-protector -Wno-unused-but-set-variable
-fomit-frame-pointer -fno-var-tracking-assignments
-Wdeclaration-after-statement -Wno-pointer-sign -fno-strict-overflow
-fconserve-stack -Werror=implicit-int -Werror=strict-prototypes
-Werror=date-time -DCC_HAVE_ASM_GOTO    -D"KBUILD_STR(s)=#s"
-D"KBUILD_BASENAME=KBUILD_STR(nommu)"
-D"KBUILD_MODNAME=KBUILD_STR(nommu)" -c -o mm/nommu.o mm/nommu.c

mm/nommu.c: In function 'delete_vma':
mm/nommu.c:861:3: error: implicit declaration of function 'vma_fput' [-Werror=implicit-function-declaration]

   vma_fput(vma);
   ^

cc1: some warnings being treated as errors

Quoted 6 lines
Quoted 1 line

Any idea what change breaks the compile?

I tried a few m68knommu defconfigs, but can't reproduce it.

Is this a plain v4.0.5? I can't find the offending call to vma_fput(). "git grep vma_fput" tells me there's no "vma_fput" in the kernel sources?

I don't have any compile (or runtime) problems with m5208evb_defconfig on linux-4.0.5 either. That is close to what most people use with qemu. What .config are you using?

Regards
Greg

Quoted 18 lines

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds
_______________________________________________
uClinux-dev mailing list
[email protected]
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by [email protected]
To unsubscribe see:
http://mailman.uclinux.org/mailman/options/uclinux-dev

#3 lore raw

Hi Greg,
Greg Ungerer wrote,

Quoted 4 lines

I don't have any compile (or runtime) problems with m5208evb_defconfig on linux-4.0.5 either. That is close to what most people use with qemu. What .config are you using?

Are you sure the defconfig creates a bootable image?

I still need at least two patches.

http://cgit.openadk.org/cgi/cgit/openadk.git/tree/target/m68k/qemu-m68k/patches/4.0.5/qemu-coldfire.patch
Without this one I get black screen.

http://cgit.openadk.org/cgi/cgit/openadk.git/tree/target/m68k/qemu-m68k/patches/4.0.5/m68k-coldfire-fec.patch
Without this one I get
qemu: hardware error: mcf_fec_read: Bad address 0x1c4
...
Abort

Networking still does not work for me after booting with the two patches applied:
fec fec.0 (unnamed net_device) (uninitialized): MDIO read timeout
fec: probe of fec.0 failed with error -5

Any idea?

best regards
Waldemar

#4 lore raw

Hi Waldemar,

Quoted 8 lines · Waldemar Brodkorb

Hi Greg,
Greg Ungerer wrote,

Quoted 4 lines

I don't have any compile (or runtime) problems with m5208evb_defconfig on linux-4.0.5 either. That is close to what most people use with qemu. What .config are you using?

Are you sure the defconfig creates a bootable image?

Yes, on real M5208EVB hardware. That is what I ran it on. I didn't run it under QEMU.

Quoted 4 lines

I still need at least two patches.

http://cgit.openadk.org/cgi/cgit/openadk.git/tree/target/m68k/qemu-m68k/patches/4.0.5/qemu-coldfire.patch
Without this one I get black screen.

Better to use the fixes to QEMU instead of working around its problems in the linux kernel sources :-)

This thread has the QEMU fixes:

https://lists.nongnu.org/archive/html/qemu-devel/2014-08/msg03224.html

I have posted them a couple of times to the QEMU dev list, but no-one seems to pick them up.

Quoted 5 lines

http://cgit.openadk.org/cgi/cgit/openadk.git/tree/target/m68k/qemu-m68k/patches/4.0.5/m68k-coldfire-fec.patch
Without this one I get
qemu: hardware error: mcf_fec_read: Bad address 0x1c4
...
Abort

Yes, you still need that. I haven't cleaned that up into a state that I am totally happy with yet.

Quoted 6 lines

Networking still does not work for me after booting with the two patches applied:
fec fec.0 (unnamed net_device) (uninitialized): MDIO read timeout
fec: probe of fec.0 failed with error -5

Any idea?

Hmm, no, haven't seen that one. I'll try a run with qemu and see if I can see what is going on.

Regards
Greg

#5 lore raw
In reply to Waldemar Brodkorb

Hi Waldemar,

Quoted 18 lines · Waldemar Brodkorb

Hi Greg,
Greg Ungerer wrote,

Quoted 3 lines

I don't have any compile (or runtime) problems with m5208evb_defconfig on linux-4.0.5 either. That is close to what most people use with qemu. What .config are you using?

Are you sure the defconfig creates a bootable image?

I still need at least two patches.

http://cgit.openadk.org/cgi/cgit/openadk.git/tree/target/m68k/qemu-m68k/patches/4.0.5/qemu-coldfire.patch
Without this one I get black screen.

http://cgit.openadk.org/cgi/cgit/openadk.git/tree/target/m68k/qemu-m68k/patches/4.0.5/m68k-coldfire-fec.patch
Without this one I get
qemu: hardware error: mcf_fec_read: Bad address 0x1c4
...
Abort

Quoted 6 lines

Networking still does not work for me after booting with the two patches applied:
fec fec.0 (unnamed net_device) (uninitialized): MDIO read timeout
fec: probe of fec.0 failed with error -5

Any idea?

The qemu fec driver does not emulate any type of attached phy or properly support the mdio read/write register.

The attached patch to qemu implements really simple phy support. Was enough for me to get linux probing and finding a phy.

Regards
Greg

Quoted 1 line

From 0a949818b081261b8322da0bef4af92f0e379655 Mon Sep 17 00:00:00 2001

From: Greg Ungerer <[email protected]>
Date: Thu, 18 Jun 2015 23:27:13 +1000
Subject: [PATCH] m68k: add simple phy support to fec driver

The Linux fec driver needs at least basic PHY support to probe and work.

This code adds a very simple set of register results for a fixed PHY setup - very similar to that used on an m5208evb board.

---
 hw/net/mcf_fec.c | 31 +++++++++++++++++++++++++++++--
 1 file changed, 29 insertions(+), 2 deletions(-)
hw/net/mcf_fec.c +29 -2
index 4bff3de..2df062f 100644
@@ -216,6 +216,33 @@ static void mcf_fec_reset(mcf_fec_state *s)
     s->rfsr = 0x500;
 }
 
+#define MMFR_WRITE_OP	(1 << 28)
+#define MMFR_READ_OP	(2 << 28)
+#define MMFR_PHYADDR(v)	(((v) >> 23) & 0x1f)
+#define MMFR_REGNUM(v)	(((v) >> 18) & 0x1f)
+
+static uint64_t mcf_fec_read_mdio(mcf_fec_state *s)
+{
+	uint64_t v;
+
+	if (s->mmfr & MMFR_WRITE_OP)
+		return s->mmfr;
+	if (MMFR_PHYADDR(s->mmfr) != 1)
+		return s->mmfr |= 0xffff;
+
+	switch (MMFR_REGNUM(s->mmfr)) {
+	case 0x00: v = 0x3100; break;
+	case 0x01: v = 0x786d; break;
+	case 0x02: v = 0x2000; break;
+	case 0x03: v = 0x5c90; break;
+	case 0x04: v = 0x01e1; break;
+	case 0x05: v = 0xc1e1; break;
+	default: v = 0xffff ; break;
+	}
+	s->mmfr = (s->mmfr & ~0xffff) | v;
+	return s->mmfr;
+}
+
 static uint64_t mcf_fec_read(void *opaque, hwaddr addr,
                              unsigned size)
 {
@@ -226,7 +253,7 @@ static uint64_t mcf_fec_read(void *opaque, hwaddr addr,
     case 0x010: return s->rx_enabled ? (1 << 24) : 0; /* RDAR */
     case 0x014: return 0; /* TDAR */
     case 0x024: return s->ecr;
-    case 0x040: return s->mmfr;
+    case 0x040: return mcf_fec_read_mdio(s);
     case 0x044: return s->mscr;
     case 0x064: return 0; /* MIBC */
     case 0x084: return s->rcr;
@@ -287,8 +314,8 @@ static void mcf_fec_write(void *opaque, hwaddr addr,
         }
         break;
     case 0x040:
-        /* TODO: Implement MII.  */
         s->mmfr = value;
+	s->eir |= FEC_INT_MII;
         break;
     case 0x044:
         s->mscr = value & 0xfe;
Signature · 1 line
1.9.1
#6 lore raw
In reply to Geert Uytterhoeven

Hi Geert,
Geert Uytterhoeven wrote,

Quoted 10 lines

Hi Waldemar,

Quoted 2 lines · Waldemar Brodkorb

I am trying to build a M68K (Coldfire no-MMU) kernel for Qemu-system-m68k. Any idea what change breaks the compile?

I tried a few m68knommu defconfigs, but can't reproduce it.

Is this a plain v4.0.5? I can't find the offending call to vma_fput(). "git grep vma_fput" tells me there's no "vma_fput" in the kernel sources?

You are totally right. It is not plain vanilla, I am really sorry about this ugly bug report. I normally use an option to disable any external kernel patches, before reporting. I missed it this time.

I have experimented with aufs4 kernel patches lately.

You are allowed to kick my ass ;)

best regards
Waldemar