Skip to contentSkip to conversation

ti-cgt6x: fix packaging issue due to libc.a

PATCH Superseded Denys Dmytriyenko 2 participants 5 messages Feb 26, 2020 – Mar 2, 2020

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ti-cgt6x: fix packaging issue due to libc.a

dwarfsrcfiles: .../work/armv7at2hf-neon-linux-gnueabi/ti-cgt6x/1_8.3.2-r0/package/usr/share/ti/cgt-c6x/lib/libc.a: not a valid ELF file

---
 recipes-ti/devtools/ti-cgt6x_8.3.2.bb | 1 +
 1 file changed, 1 insertion(+)
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +1 -0
index c63c3f6..1188b4e 100644
@@ -100,6 +100,7 @@ FILES_${PN} += "${datadir}/ti/*"
 FILES_${PN}-dbg += "${datadir}/ti/cgt-c6x/bin/.debug"
 
 INSANE_SKIP_${PN} += "staticdev"
+INHIBIT_PACKAGE_DEBUG_SPLIT = "1"
 
 BBCLASSEXTEND = "native nativesdk"
 
Signature · 1 line
2.7.4
#1 lore raw
Quoted 12 lines · Denys Dmytriyenko via Lists.Yoctoproject.Org

dwarfsrcfiles: .../work/armv7at2hf-neon-linux-gnueabi/ti-cgt6x/1_8.3.2-r0/package/usr/share/ti/cgt-c6x/lib/libc.a: not a valid ELF file

---
 recipes-ti/devtools/ti-cgt6x_8.3.2.bb | 1 +
 1 file changed, 1 insertion(+)
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +0 -0
index c63c3f6..1188b4e 100644

recipes-ti/devtools/ti-cgt6x_8.3.2.bb FILES_${PN} += "${datadir}/ti/*" · line 104

Quoted 5 lines
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +1 -0
@@ -100,6 +100,7 @@ FILES_${PN} += "${datadir}/ti/*"
 FILES_${PN}-dbg += "${datadir}/ti/cgt-c6x/bin/.debug"
 
 INSANE_SKIP_${PN} += "staticdev"
+INHIBIT_PACKAGE_DEBUG_SPLIT = "1"
 

Thanks for taking care of this issue, does it have any downsides, like not stripping binaries which otherwise should have been stripped besides libc.a

Quoted 6 lines
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +0 -0
 BBCLASSEXTEND = "native nativesdk"
 
#2 lore raw
Quoted 17 lines · Khem Raj
Quoted 14 lines · Denys Dmytriyenko via Lists.Yoctoproject.Org

dwarfsrcfiles: .../work/armv7at2hf-neon-linux-gnueabi/ti-cgt6x/1_8.3.2-r0/package/usr/share/ti/cgt-c6x/lib/libc.a: not a valid ELF file

---
 recipes-ti/devtools/ti-cgt6x_8.3.2.bb | 1 +
 1 file changed, 1 insertion(+)
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +0 -0
index c63c3f6..1188b4e 100644
@@ -100,6 +100,7 @@ FILES_${PN} += "${datadir}/ti/*"
 FILES_${PN}-dbg += "${datadir}/ti/cgt-c6x/bin/.debug"
 
Quoted 7 lines

recipes-ti/devtools/ti-cgt6x_8.3.2.bb FILES_${PN} += "${datadir}/ti/*" · line 104

Quoted 3 lines
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +1 -0
 INSANE_SKIP_${PN} += "staticdev"
+INHIBIT_PACKAGE_DEBUG_SPLIT = "1"
 

Thanks for taking care of this issue, does it have any downsides, like not stripping binaries which otherwise should have been stripped besides libc.a

This is C6000 DSP code and is not the main architecture for the platform being built, so standard ARM tools won't be able to handle that anyway. But all the other ARM binaries in that package are not stripped now, unfortunately.

On the other hand, I'm getting the same error from dwarfsrcfiles on another static .a library from another package, which contains the proper ARM code, which is rather strange. The only suspicious thing there is an empty metafile called ".created", but for now I have to disable stripping there as well...

Quoted 6 lines
Quoted 6 lines
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +0 -0
 BBCLASSEXTEND = "native nativesdk"
 
#3 lore raw
Quoted 26 lines · Denys Dmytriyenko
Quoted 17 lines · Khem Raj
Quoted 14 lines · Denys Dmytriyenko via Lists.Yoctoproject.Org

dwarfsrcfiles: .../work/armv7at2hf-neon-linux-gnueabi/ti-cgt6x/1_8.3.2-r0/package/usr/share/ti/cgt-c6x/lib/libc.a: not a valid ELF file

---
 recipes-ti/devtools/ti-cgt6x_8.3.2.bb | 1 +
 1 file changed, 1 insertion(+)
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +0 -0
index c63c3f6..1188b4e 100644
@@ -100,6 +100,7 @@ FILES_${PN} += "${datadir}/ti/*"
 FILES_${PN}-dbg += "${datadir}/ti/cgt-c6x/bin/.debug"
 
Quoted 7 lines

recipes-ti/devtools/ti-cgt6x_8.3.2.bb FILES_${PN} += "${datadir}/ti/*" · line 104

Quoted 3 lines
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +1 -0
 INSANE_SKIP_${PN} += "staticdev"
+INHIBIT_PACKAGE_DEBUG_SPLIT = "1"
 

Thanks for taking care of this issue, does it have any downsides, like not stripping binaries which otherwise should have been stripped besides libc.a

Quoted 4 lines

This is C6000 DSP code and is not the main architecture for the platform being built, so standard ARM tools won't be able to handle that anyway. But all the other ARM binaries in that package are not stripped now, unfortunately.

yeah I thought so, and dwarfsrcfiles is quite simplistic, perhaps you might want to file a defect so we could define a WHITELIST of files in recipes which should be spared from
stripping

Quoted 12 lines

On the other hand, I'm getting the same error from dwarfsrcfiles on another static .a library from another package, which contains the proper ARM code, which is rather strange. The only suspicious thing there is an empty metafile called ".created", but for now I have to disable stripping there as well...

Quoted 6 lines
Quoted 6 lines
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +0 -0
 BBCLASSEXTEND = "native nativesdk"
 
#4 lore raw
Quoted 29 lines · Khem Raj
Quoted 26 lines · Denys Dmytriyenko
Quoted 24 lines · Khem Raj
Quoted 12 lines · Denys Dmytriyenko via Lists.Yoctoproject.Org

dwarfsrcfiles: .../work/armv7at2hf-neon-linux-gnueabi/ti-cgt6x/1_8.3.2-r0/package/usr/share/ti/cgt-c6x/lib/libc.a: not a valid ELF file

---
 recipes-ti/devtools/ti-cgt6x_8.3.2.bb | 1 +
 1 file changed, 1 insertion(+)
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +0 -0
index c63c3f6..1188b4e 100644

recipes-ti/devtools/ti-cgt6x_8.3.2.bb FILES_${PN} += "${datadir}/ti/*" · line 104

Quoted 5 lines
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +1 -0
@@ -100,6 +100,7 @@ FILES_${PN} += "${datadir}/ti/*"
 FILES_${PN}-dbg += "${datadir}/ti/cgt-c6x/bin/.debug"
 
 INSANE_SKIP_${PN} += "staticdev"
+INHIBIT_PACKAGE_DEBUG_SPLIT = "1"
 

Thanks for taking care of this issue, does it have any downsides, like not stripping binaries which otherwise should have been stripped besides libc.a

Quoted 8 lines
Quoted 3 lines

This is C6000 DSP code and is not the main architecture for the platform being built, so standard ARM tools won't be able to handle that anyway. But all the other ARM binaries in that package are not stripped now, unfortunately.

yeah I thought so, and dwarfsrcfiles is quite simplistic, perhaps you might want to file a defect so we could define a WHITELIST of files in recipes which should be spared from
stripping

Sent v2 to use INHIBIT_PACKAGE_STRIP_FILES for whitelisting.

Quoted 12 lines
Quoted 12 lines

On the other hand, I'm getting the same error from dwarfsrcfiles on another static .a library from another package, which contains the proper ARM code, which is rather strange. The only suspicious thing there is an empty metafile called ".created", but for now I have to disable stripping there as well...

Quoted 6 lines
Quoted 6 lines
recipes-ti/devtools/ti-cgt6x_8.3.2.bb +0 -0
 BBCLASSEXTEND = "native nativesdk"