Fix internal blackfin toolchain

PATCH New Gustavo Zacarias 3 participants 20 messages Jul 20, 2013 – Jul 27, 2013

Share & tools

Gustavo Zacarias

author

2013-07-20 12:52 UTC #0 lore raw

cover Fix internal blackfin toolchain

This patchset fixes the internal blackfin toolchain build. It's been broken for ages (never worked maybe?)
Testers wanted, it builds what's supposed to build but i lack hardware to see if it really works.

Gustavo Zacarias (9):
gcc/bfin: doesn't support --with-cpu defaults uclibc/bfin: only supports linuxthreads.old
uclibc: make target utils optional
gcc/bfin: 4.{6,7,8}.x are broken
uclibc: use builtin startfiles target uclibc/bfin: set the appropiate binfmt in the configuration uclibc/bfin: don't install host utils if not FDPIC
blackfin: set GNU_TARGET_NAME accordingly gcc/mudflap: disable for blackfin, it doesn't build

 package/Makefile.in        |  5 +++
 package/gcc/Config.in.host |  7 +++--
 package/gcc/gcc.mk         |  8 +++++
 package/uclibc/Config.in   | 15 ++++++++-
 package/uclibc/uclibc.mk   | 76 ++++++++++++++++++++++++++++++++++++++--------
 5 files changed, 95 insertions(+), 16 deletions(-)

--
1.8.1.5

Gustavo Zacarias

author

2013-07-20 12:52 UTC #1 lore raw

PATCH 1/9 gcc/bfin: doesn't support --with-cpu defaults

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/gcc/gcc.mk | 3 +++
 1 file changed, 3 insertions(+)

package/gcc/gcc.mk +3 -0

index 85cd7d4..cbeaa6c 100644
@@ -159,6 +159,8 @@ endif
 ifneq ($(call qstrip,$(BR2_GCC_TARGET_ABI)),)
 HOST_GCC_COMMON_CONF_OPT += --with-abi=$(BR2_GCC_TARGET_ABI)
 endif
+# GCC doesn't support --with-cpu for bfin
+ifeq ($(BR2_bfin),)
 ifneq ($(call qstrip,$(BR2_GCC_TARGET_CPU)),)
 ifneq ($(call qstrip,$(BR2_GCC_TARGET_CPU_REVISION)),)
 HOST_GCC_COMMON_CONF_OPT += --with-cpu=$(call qstrip,$(BR2_GCC_TARGET_CPU)-$(BR2_GCC_TARGET_CPU_REVISION))

@@ -166,6 +168,7 @@ else
 HOST_GCC_COMMON_CONF_OPT += --with-cpu=$(call qstrip,$(BR2_GCC_TARGET_CPU))
 endif
 endif
+endif
 
 GCC_TARGET_FPU = $(call qstrip,$(BR2_GCC_TARGET_FPU))
 ifneq ($(GCC_TARGET_FPU),)

Signature · 1 line

1.8.1.5

Thomas Petazzoni

2013-07-21 15:35 UTC #2 lore raw

PATCH 1/9 gcc/bfin: doesn't support --with-cpu defaults

Dear Gustavo Zacarias,

package/gcc/gcc.mk else · line 171

Quoted 12 lines · Gustavo Zacarias

Signed-off-by Gustavo Zacarias <[email protected]>

package/gcc/gcc.mk +3 -0

+# GCC doesn't support --with-cpu for bfin
+ifeq ($(BR2_bfin),)
 ifneq ($(call qstrip,$(BR2_GCC_TARGET_CPU)),)
 ifneq ($(call qstrip,$(BR2_GCC_TARGET_CPU_REVISION)),)
 HOST_GCC_COMMON_CONF_OPT += --with-cpu=$(call qstrip,$(BR2_GCC_TARGET_CPU)-$(BR2_GCC_TARGET_CPU_REVISION))

@@ -166,6 +168,7 @@ else
 HOST_GCC_COMMON_CONF_OPT += --with-cpu=$(call qstrip,$(BR2_GCC_TARGET_CPU))
 endif
 endif
+endif

This looks odd because the BR2_GCC_TARGET_CPU_REVISION thing was added by Sonic Zhang, see 66d41890ec2b76189bcd427a0cc3966ff56f9712. So it is specifically useful on Blackfin. Maybe just with Analog Devices toolchains, I don't know. But clearly, removing that for Blackfin seems wrong.

Signature · 4 lines

Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

Gustavo Zacarias

author

2013-07-21 16:26 UTC #3 lore raw

PATCH 1/9 gcc/bfin: doesn't support --with-cpu defaults

Quoted 5 lines · Thomas Petazzoni

This looks odd because the BR2_GCC_TARGET_CPU_REVISION thing was added by Sonic Zhang, see 66d41890ec2b76189bcd427a0cc3966ff56f9712. So it is specifically useful on Blackfin. Maybe just with Analog Devices toolchains, I don't know. But clearly, removing that for Blackfin seems wrong.

Exactly, with their twisted toolchains, snip without it:

checking for clock_t... yes
checking for .preinit_array/.init_array/.fini_array support... yes checking if mkdir takes one argument... no
This target does not support --with-cpu.
Valid --with options are:
make[1]: *** [configure-gcc] Error 1
make[1]: Leaving directory `/home/gustavoz/b/fin/output/build/host-gcc-initial-4.5.4/build'
make: *** [/home/gustavoz/b/fin/output/build/host-gcc-initial-4.5.4/.stamp_built] Error 2

Regards.

Gustavo Zacarias

author

2013-07-20 12:52 UTC #4 lore raw

PATCH 2/9 uclibc/bfin: only supports linuxthreads.old

In reply to Gustavo Zacarias

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/uclibc/Config.in | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

package/uclibc/Config.in +3 -1

index 90aae90..f9c4956 100644
@@ -104,6 +104,7 @@ choice
 
 	config BR2_PTHREADS
 		bool "linuxthreads"
+		depends on !BR2_bfin
 		depends on !BR2_xtensa
 	        select BR2_TOOLCHAIN_HAS_THREADS

@@ -115,9 +116,10 @@ choice
 		bool "Native POSIX Threading (NPTL)"
 		select BR2_TOOLCHAIN_HAS_THREADS
 		depends on !BR2_x86_i386
+		depends on !BR2_arc
 		depends on !BR2_avr32
+		depends on !BR2_bfin
 		depends on !BR2_xtensa
-		depends on !BR2_arc
 endchoice
 
 config BR2_PTHREAD_DEBUG

Signature · 1 line

1.8.1.5

Yann E. MORIN

2013-07-20 12:58 UTC #5 lore raw

PATCH 2/9 uclibc/bfin: only supports linuxthreads.old

Gustavo, All,

On 2013-07-20 09:52 -0300, Gustavo Zacarias spake thusly:

Quoted 18 lines

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/uclibc/Config.in | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

package/uclibc/Config.in +1 -0

index 90aae90..f9c4956 100644
@@ -104,6 +104,7 @@ choice
 
 	config BR2_PTHREADS
 		bool "linuxthreads"
+		depends on !BR2_bfin
 		depends on !BR2_xtensa
 	        select BR2_TOOLCHAIN_HAS_THREADS

package/uclibc/Config.in choice · line 120

Quoted 8 lines

package/uclibc/Config.in +2 -1

@@ -115,9 +116,10 @@ choice
 		bool "Native POSIX Threading (NPTL)"
 		select BR2_TOOLCHAIN_HAS_THREADS
 		depends on !BR2_x86_i386
+		depends on !BR2_arc
 		depends on !BR2_avr32
+		depends on !BR2_bfin
 		depends on !BR2_xtensa
-		depends on !BR2_arc

While at re-ordering, you could have also reorderd x86_i386. Note this in the commit log, such as:
trivial options re-rodering at the same time.

Although I would prefer you not mix cosmetics with actual changes.

Regards,
Yann E. MORIN.

Signature · 6 lines

.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

Thomas Petazzoni

2013-07-27 14:21 UTC #6 lore raw

PATCH 2/9 uclibc/bfin: only supports linuxthreads.old

In reply to Gustavo Zacarias

Dear Gustavo Zacarias,

Quoted 18 lines · Gustavo Zacarias

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/uclibc/Config.in | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

package/uclibc/Config.in +1 -0

index 90aae90..f9c4956 100644
@@ -104,6 +104,7 @@ choice
 
 	config BR2_PTHREADS
 		bool "linuxthreads"
+		depends on !BR2_bfin
 		depends on !BR2_xtensa
 	        select BR2_TOOLCHAIN_HAS_THREADS

package/uclibc/Config.in choice · line 123

Quoted 9 lines

package/uclibc/Config.in +2 -1

@@ -115,9 +116,10 @@ choice
 		bool "Native POSIX Threading (NPTL)"
 		select BR2_TOOLCHAIN_HAS_THREADS
 		depends on !BR2_x86_i386
+		depends on !BR2_arc
 		depends on !BR2_avr32
+		depends on !BR2_bfin
 		depends on !BR2_xtensa
-		depends on !BR2_arc
 endchoice

As suggested by Yann, I've splitted this in two patches, one adding the !bfin dependency, another one doing the re-ordering, including x86_i386.

Thomas

Signature · 4 lines

Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

Gustavo Zacarias

author

2013-07-20 12:52 UTC #7 lore raw

PATCH 3/9 uclibc: make target utils optional

In reply to Gustavo Zacarias

These are broken for blackfin unfortunately so they're disabled.

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/uclibc/Config.in | 11 +++++++++++
 package/uclibc/uclibc.mk | 16 +++++++++++-----
 2 files changed, 22 insertions(+), 5 deletions(-)

package/uclibc/Config.in +11 -0

index f9c4956..32289bb 100644
@@ -129,6 +129,17 @@ config BR2_PTHREAD_DEBUG
 	help
 	  Build the thread library with debugging enabled.
 
+config BR2_UCLIBC_INSTALL_UTILS
+	bool "Compile and install uClibc utilities"
+	depends on !BR2_bfin
+	default y
+	help
+	  Enabling this option will compile and install the getconf,
+	  ldconfig and ldd uClibc utilities for the target.
+
+	  You can save ~32 KiB in target space by disabling them since
+	  they're normally not needed.
+
 config BR2_UCLIBC_INSTALL_TEST_SUITE
 	bool "Compile and install uClibc tests"
 	select BR2_PACKAGE_MAKE

package/uclibc/uclibc.mk +11 -5

index cf53280..4a37fa0 100644
@@ -451,6 +451,16 @@ define UCLIBC_INSTALL_TEST_SUITE
 endef
 endif
 
+ifeq ($(BR2_UCLIBC_INSTALL_UTILS),y)
+define UCLIBC_INSTALL_UTILS_TARGET
+	$(MAKE1) -C $(@D) \
+		CC="$(TARGET_CC)" CPP="$(TARGET_CPP)" LD="$(TARGET_LD)" \
+		ARCH="$(UCLIBC_TARGET_ARCH)" \
+		PREFIX=$(TARGET_DIR) \
+		utils install_utils
+endef
+endif
+
 define UCLIBC_INSTALL_TARGET_CMDS
 	$(MAKE1) -C $(@D) \
 		$(UCLIBC_MAKE_FLAGS) \

@@ -458,11 +468,7 @@ define UCLIBC_INSTALL_TARGET_CMDS
 		DEVEL_PREFIX=/usr/ \
 		RUNTIME_PREFIX=/ \
 		install_runtime
-	$(MAKE1) -C $(@D) \
-		CC="$(TARGET_CC)" CPP="$(TARGET_CPP)" LD="$(TARGET_LD)" \
-		ARCH="$(UCLIBC_TARGET_ARCH)" \
-		PREFIX=$(TARGET_DIR) \
-		utils install_utils
+	$(UCLIBC_INSTALL_UTILS_TARGET)
 	$(UCLIBC_INSTALL_TEST_SUITE)
 endef

Signature · 1 line

1.8.1.5

Yann E. MORIN

2013-07-20 13:08 UTC #8 lore raw

PATCH 3/9 uclibc: make target utils optional

On 2013-07-20 09:52 -0300, Gustavo Zacarias spake thusly:

Quoted 3 lines

These are broken for blackfin unfortunately so they're disabled.
Signed-off-by Gustavo Zacarias <[email protected]>

Reviewed-by "Yann E. MORIN" <[email protected]>

Acked-by "Yann E. MORIN" <[email protected]>

BTW, I never know if Acked-by includes Reviewed-by...

Regads,
Yann E. MORIN.

Signature · 6 lines

.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 223 225 172 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'

Gustavo Zacarias

author

2013-07-20 12:52 UTC #9 lore raw

PATCH 4/9 gcc/bfin: 4.{6,7,8}.x are broken

In reply to Gustavo Zacarias

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/gcc/Config.in.host | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

package/gcc/Config.in.host +4 -3

index c49b8ed..a6acaa2 100644
@@ -8,6 +8,7 @@ choice
 	default BR2_GCC_VERSION_4_4_X if BR2_sparc_sparchfleon || BR2_sparc_sparchfleonv8 || BR2_sparc_sparcsfleon || BR2_sparc_sparcsfleonv8
 	default BR2_GCC_VERSION_4_2_2_AVR32_2_1_5 if BR2_avr32
 	default BR2_GCC_VERSION_4_8_ARC if BR2_arc
+	default BR2_GCC_VERSION_4_5_X if BR2_bfin
 	default BR2_GCC_VERSION_4_7_X
 	help
 	  Select the version of gcc you wish to use.

@@ -37,17 +38,17 @@ choice
 		bool "gcc 4.5.x"
 
 	config BR2_GCC_VERSION_4_6_X
-		depends on !BR2_arc && !BR2_avr32 && !BR2_cortex_a7 && !BR2_sparc_sparchfleon && !BR2_sparc_sparchfleonv8 && !BR2_sparc_sparcsfleon && !BR2_sparc_sparcsfleonv8 && !BR2_pj4
+		depends on !BR2_arc && !BR2_avr32 && !BR2_bfin && !BR2_cortex_a7 && !BR2_sparc_sparchfleon && !BR2_sparc_sparchfleonv8 && !BR2_sparc_sparcsfleon && !BR2_sparc_sparcsfleonv8 && !BR2_pj4
 		select BR2_GCC_NEEDS_MPC
 		bool "gcc 4.6.x"
 
 	config BR2_GCC_VERSION_4_7_X
-		depends on !BR2_arc && !BR2_avr32 && !BR2_sparc_sparchfleon && !BR2_sparc_sparchfleonv8 && !BR2_sparc_sparcsfleon && !BR2_sparc_sparcsfleonv8 && !BR2_pj4
+		depends on !BR2_arc && !BR2_avr32 && !BR2_bfin && !BR2_sparc_sparchfleon && !BR2_sparc_sparchfleonv8 && !BR2_sparc_sparcsfleon && !BR2_sparc_sparcsfleonv8 && !BR2_pj4
 		select BR2_GCC_NEEDS_MPC
 		bool "gcc 4.7.x"
 
 	config BR2_GCC_VERSION_4_8_X
-		depends on !BR2_arc && !BR2_avr32 && !BR2_sparc_sparchfleon && !BR2_sparc_sparchfleonv8 && !BR2_sparc_sparcsfleon && !BR2_sparc_sparcsfleonv8
+		depends on !BR2_arc && !BR2_avr32 && !BR2_bfin && !BR2_sparc_sparchfleon && !BR2_sparc_sparchfleonv8 && !BR2_sparc_sparcsfleon && !BR2_sparc_sparcsfleonv8
 		select BR2_GCC_NEEDS_MPC
 		bool "gcc 4.8.x"

Signature · 1 line

1.8.1.5

Gustavo Zacarias

author

2013-07-20 12:52 UTC #10 lore raw

PATCH 5/9 uclibc: use builtin startfiles target

In reply to Gustavo Zacarias

Now that we've dropped 0.9.31.x we can use the startfiles / install_startfiles targets instead of building and installing them by hand.
Fixes internal blackfin toolchain build errors regarding crtreloc.o

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/uclibc/uclibc.mk | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

package/uclibc/uclibc.mk +2 -3

index 4a37fa0..efe00a6 100644
@@ -409,13 +409,12 @@ define UCLIBC_CONFIGURE_CMDS
 		PREFIX=$(STAGING_DIR) \
 		DEVEL_PREFIX=/usr/ \
 		RUNTIME_PREFIX=$(STAGING_DIR) \
-		headers lib/crt1.o lib/crti.o lib/crtn.o \
-		install_headers
+		headers startfiles \
+		install_headers install_startfiles
 	$(TARGET_CROSS)gcc -nostdlib \
 		-nostartfiles -shared -x c /dev/null -o $(STAGING_DIR)/usr/lib/libc.so
 	$(TARGET_CROSS)gcc -nostdlib \
 		-nostartfiles -shared -x c /dev/null -o $(STAGING_DIR)/usr/lib/libm.so
-	cp -pLR $(UCLIBC_DIR)/lib/crt[1in].o $(STAGING_DIR)/usr/lib/
 endef
 
 ifeq ($(BR2_UCLIBC_INSTALL_TEST_SUITE),y)

Signature · 1 line

1.8.1.5

Gustavo Zacarias

author

2013-07-20 12:52 UTC #11 lore raw

PATCH 6/9 uclibc/bfin: set the appropiate binfmt in the configuration

In reply to Gustavo Zacarias

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/uclibc/uclibc.mk | 40 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)

package/uclibc/uclibc.mk +40 -0

index efe00a6..cbc2e56 100644
@@ -148,6 +148,45 @@ endef
 endif # powerpc
 
 #
+# Blackfin definitions
+#
+
+ifeq ($(UCLIBC_TARGET_ARCH),bfin)
+ifeq ($(BR2_BINFMT_FDPIC),y)
+define UCLIBC_BFIN_CONFIG
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_FLAT,$(@D))
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_FLAT_SEP_DATA,$(@D))
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_SHARED_FLAT,$(@D))
+	$(call UCLIBC_OPT_SET,UCLIBC_FORMAT_FDPIC_ELF,y,$(@D))
+endef
+endif
+ifeq ($(BR2_BINFMT_FLAT_ONE),y)
+define UCLIBC_BFIN_CONFIG
+	$(call UCLIBC_OPT_SET,UCLIBC_FORMAT_FLAT,y,$(@D))
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_FLAT_SEP_DATA,$(@D))
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_SHARED_FLAT,$(@D))
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_FDPIC_ELF,$(@D))
+endef
+endif
+ifeq ($(BR2_BINFMT_FLAT_SEP_DATA),y)
+define UCLIBC_BFIN_CONFIG
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_FLAT,$(@D))
+	$(call UCLIBC_OPT_SET,UCLIBC_FORMAT_FLAT_SEP_DATA,y,$(@D))
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_SHARED_FLAT,$(@D))
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_FDPIC_ELF,$(@D))
+endef
+endif
+ifeq ($(BR2_BINFMT_FLAT_SHARED),y)
+define UCLIBC_BFIN_CONFIG
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_FLAT,$(@D))
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_FLAT_SEP_DATA,$(@D))
+	$(call UCLIBC_OPT_SET,UCLIBC_FORMAT_SHARED_FLAT,y,$(@D))
+	$(call UCLIBC_OPT_UNSET,UCLIBC_FORMAT_FDPIC_ELF,$(@D))
+endef
+endif
+endif # bfin
+
+#
 # AVR32 definitions
 #

@@ -380,6 +419,7 @@ define UCLIBC_SETUP_DOT_CONFIG
 	$(UCLIBC_SPARC_TYPE_CONFIG)
 	$(UCLIBC_POWERPC_TYPE_CONFIG)
 	$(UCLIBC_AVR32_CONFIG)
+	$(UCLIBC_BFIN_CONFIG)
 	$(UCLIBC_X86_TYPE_CONFIG)
 	$(UCLIBC_ENDIAN_CONFIG)
 	$(UCLIBC_LARGEFILE_CONFIG)

Signature · 1 line

1.8.1.5

Gustavo Zacarias

author

2013-07-20 12:52 UTC #12 lore raw

PATCH 7/9 uclibc/bfin: don't install host utils if not FDPIC

In reply to Gustavo Zacarias

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/uclibc/uclibc.mk | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

package/uclibc/uclibc.mk +11 -4

index cbc2e56..644c531 100644
@@ -511,6 +511,16 @@ define UCLIBC_INSTALL_TARGET_CMDS
 	$(UCLIBC_INSTALL_TEST_SUITE)
 endef
 
+# For FLAT binfmts (static) there are no host utils
+ifeq ($(BR2_BINFMT_FLAT),)
+define UCLIBC_INSTALL_UTILS_STAGING
+	$(INSTALL) -D -m 0755 $(@D)/utils/ldd.host $(HOST_DIR)/usr/bin/ldd
+	ln -sf ldd $(HOST_DIR)/usr/bin/$(GNU_TARGET_NAME)-ldd
+	$(INSTALL) -D -m 0755 $(@D)/utils/ldconfig.host $(HOST_DIR)/usr/bin/ldconfig
+	ln -sf ldconfig $(HOST_DIR)/usr/bin/$(GNU_TARGET_NAME)-ldconfig
+endef
+endif
+
 define UCLIBC_INSTALL_STAGING_CMDS
 	$(MAKE1) -C $(@D) \
 		$(UCLIBC_MAKE_FLAGS) \

@@ -518,10 +528,7 @@ define UCLIBC_INSTALL_STAGING_CMDS
 		DEVEL_PREFIX=/usr/ \
 		RUNTIME_PREFIX=/ \
 		install_runtime install_dev
-	install -D -m 0755 $(@D)/utils/ldd.host $(HOST_DIR)/usr/bin/ldd
-	ln -sf ldd $(HOST_DIR)/usr/bin/$(GNU_TARGET_NAME)-ldd
-	install -D -m 0755 $(@D)/utils/ldconfig.host $(HOST_DIR)/usr/bin/ldconfig
-	ln -sf ldconfig $(HOST_DIR)/usr/bin/$(GNU_TARGET_NAME)-ldconfig
+	$(UCLIBC_INSTALL_UTILS_STAGING)
 endef
 
 uclibc-menuconfig: dirs uclibc-patch

Signature · 1 line

1.8.1.5

Gustavo Zacarias

author

2013-07-20 12:52 UTC #13 lore raw

PATCH 8/9 blackfin: set GNU_TARGET_NAME accordingly

In reply to Gustavo Zacarias

Blackfin for FDPIC targets uses *-linux-*, however for FLAT targets it needs *-uclinux-* in order for gcc to build properly.

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/Makefile.in | 5 +++++
 1 file changed, 5 insertions(+)

package/Makefile.in +5 -0

index a23eb41..acbc453 100644
@@ -23,6 +23,11 @@ MAKE:=$(HOSTMAKE) -j$(PARALLEL_JOBS)
 # Compute GNU_TARGET_NAME
 GNU_TARGET_NAME=$(ARCH)-buildroot-linux-$(LIBC)$(ABI)
 
+# Blackfin FLAT needs uclinux
+ifeq ($(BR2_bfin)$(BR2_BINFMT_FLAT),yy)
+GNU_TARGET_NAME=$(ARCH)-buildroot-uclinux-$(LIBC)$(ABI)
+endif
+
 ifeq ($(BR2_TOOLCHAIN_USES_UCLIBC),y)
 LIBC=uclibc
 else

Signature · 1 line

1.8.1.5

Thomas Petazzoni

2013-07-21 15:38 UTC #14 lore raw

PATCH 8/9 blackfin: set GNU_TARGET_NAME accordingly

Dear Gustavo Zacarias,

package/Makefile.in MAKE:=$(HOSTMAKE) -j$(PARALLEL_JOBS) · line 29

Quoted 8 lines · Gustavo Zacarias

package/Makefile.in +4 -0

@@ -23,6 +23,11 @@ MAKE:=$(HOSTMAKE) -j$(PARALLEL_JOBS)
 # Compute GNU_TARGET_NAME
 GNU_TARGET_NAME=$(ARCH)-buildroot-linux-$(LIBC)$(ABI)
 
+# Blackfin FLAT needs uclinux
+ifeq ($(BR2_bfin)$(BR2_BINFMT_FLAT),yy)
+GNU_TARGET_NAME=$(ARCH)-buildroot-uclinux-$(LIBC)$(ABI)
+endif

Isn't the 'uclinux' OS part something that is not specific to Blackfin? I.e, should this rather be something like:

ifeq ($(BR2_USE_MMU),y)
OS=linux
else
OS=uclinux
endif

GNU_TARGET_NAME=$(ARCH)-buildroot-$(OS)-$(LIBC)$(ABI)

Thanks!

Thomas

Signature · 4 lines

Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

Gustavo Zacarias

author

2013-07-21 16:32 UTC #15 lore raw

PATCH 8/9 blackfin: set GNU_TARGET_NAME accordingly

Quoted 8 lines · Thomas Petazzoni

Isn't the 'uclinux' OS part something that is not specific to Blackfin? I.e, should this rather be something like:
ifeq ($(BR2_USE_MMU),y)
OS=linux
else
OS=uclinux
endif

In theory, yes.
For blackfin, doesn't seem so.
Seems (caveat: no access to blackfin.uclinux.org IF it's even documented, bizarre patches and so on) that for blackfin being MMU-less across the whole arch it's used to select from FDPIC and FLAT*. Again going to the source (gcc to be precise) we see that crtreloc.o (for FDPIC targets) is used in gcc/config/bfin/linux.h (linux tuple), and crtlibid (for FLAT targets, internal to gcc) is used in gcc/config/bfin/uclinux.h
crtreloc.o is generated by uClibc only when doing FDPIC, and used by 'linux', and 'uclinux' doesn't do FDPIC (no linking with crtreloc.o) hence FLAT.
Regards.

Thomas Petazzoni

2013-07-27 14:22 UTC #16 lore raw

PATCH 8/9 blackfin: set GNU_TARGET_NAME accordingly

In reply to Gustavo Zacarias

Dear Gustavo Zacarias,

package/Makefile.in MAKE:=$(HOSTMAKE) -j$(PARALLEL_JOBS) · line 29

Quoted 9 lines · Gustavo Zacarias

Blackfin for FDPIC targets uses *-linux-*, however for FLAT targets it
# Compute GNU_TARGET_NAME GNU_TARGET_NAME=$(ARCH)-buildroot-linux-$(LIBC)$(ABI)
package/Makefile.in +4 -0
+# Blackfin FLAT needs uclinux
+ifeq ($(BR2_bfin)$(BR2_BINFMT_FLAT),yy)
+GNU_TARGET_NAME=$(ARCH)-buildroot-uclinux-$(LIBC)$(ABI)
+endif

I've changed that to keep a single GNU_TARGET_NAME definition and to use a new TARGET_OS variable, like the existing LIBC and ABI variables we had.

Thomas

Signature · 4 lines

Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

Gustavo Zacarias

author

2013-07-20 12:52 UTC #17 lore raw

PATCH 9/9 gcc/mudflap: disable for blackfin, it doesn't build

In reply to Gustavo Zacarias

Signed-off-by Gustavo Zacarias <[email protected]>

---
 package/gcc/gcc.mk | 5 +++++
 1 file changed, 5 insertions(+)

package/gcc/gcc.mk +5 -0

index cbeaa6c..cd6efed 100644
@@ -202,6 +202,11 @@ ifeq ($(BR2_ARM_INSTRUCTIONS_THUMB),y)
 HOST_GCC_COMMON_CONF_OPT += --disable-libmudflap
 endif
 
+# Blackfin doesn't do mudflap
+ifeq ($(BR2_bfin),y)
+HOST_GCC_COMMON_CONF_OPT += --disable-libmudflap
+endif
+
 # Disable mudflap and enable proper double/long double for SPE ABI
 ifeq ($(BR2_powerpc_SPE),y)
 HOST_GCC_COMMON_CONF_OPT += \

Signature · 1 line

1.8.1.5

Thomas Petazzoni

2013-07-27 14:23 UTC #18 lore raw

cover Fix internal blackfin toolchain

In reply to Gustavo Zacarias

Dear Gustavo Zacarias,

Quoted 5 lines · Gustavo Zacarias

This patchset fixes the internal blackfin toolchain build. It's been broken for ages (never worked maybe?)
Testers wanted, it builds what's supposed to build but i lack hardware to see if it really works.

Quoted 10 lines

Gustavo Zacarias (9):
gcc/bfin: doesn't support --with-cpu defaults uclibc/bfin: only supports linuxthreads.old
uclibc: make target utils optional
gcc/bfin: 4.{6,7,8}.x are broken
uclibc: use builtin startfiles target uclibc/bfin: set the appropiate binfmt in the configuration uclibc/bfin: don't install host utils if not FDPIC
blackfin: set GNU_TARGET_NAME accordingly gcc/mudflap: disable for blackfin, it doesn't build

Committed entire series, with the minor fixes I pointed on the individual patches.

Note that when you select FLAT format, it is still ELF binaries that are produced in $(O)/target. Most likely related to our non-working handling of elf2flt, as discussed recently on the list regarding ARM noMMU platforms. So this patch series certainly improves the situation, but doesn't fix the whole thing yet :)

Thanks!

Thomas

Signature · 4 lines

Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

Gustavo Zacarias

author

2013-07-27 14:35 UTC #19 lore raw

cover Fix internal blackfin toolchain

Quoted 8 lines · Thomas Petazzoni

Committed entire series, with the minor fixes I pointed on the individual patches.
Note that when you select FLAT format, it is still ELF binaries that are produced in $(O)/target. Most likely related to our non-working handling of elf2flt, as discussed recently on the list regarding ARM noMMU platforms. So this patch series certainly improves the situation, but doesn't fix the whole thing yet :)

Thanks.
Yes, as i've stated, it's a first patchset/try of fixing things, but most certainly not tested :)
I've been playing a bit with fixing ARM nommu, quick notes say we need to ditch libmudflap, change the tuple (uclinux-eabi & maybe uclinux-elf), make host-elf2flt a hard dep (it's invoked at link time by gcc with the appropiate tuple), probably add the mmu-less cortex processors (M*, R4), and get someone to test the whole thing. I've still didn't get the whole thing to build properly but it's close. Regards.