TI tools - make EULA unpack work when /opt is writeable

PATCH In review Gary Thomas 5 participants 7 messages Aug 28, 2010 – Nov 16, 2010

Share & tools

Gary Thomas

author

2010-08-28 09:08 UTC #0 lore raw

PATCH v2 1/1 TI tools - make EULA unpack work when /opt is writeable

This patch addresses the problem where the TI tools that need the EULA unpack into the wrong place if the user has write permission in /opt. The current behaviour of expecting the unpacker to fall back to the $HOME environment variable should not be relied on. The patch changes the unpack step to use the --prefix= option which will always work.

Version 2:
* Update patch to be against org.openembedded.dev branch
* Incorporate full description into patch

Signature · 4 lines

------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

From a5f48f6984c2f075493ca3b44b437ab9aa94ca02 Mon Sep 17 00:00:00 2001
From: Gary Thomas <[email protected]>
Date: Sat, 28 Aug 2010 02:55:54 -0600
Subject: [PATCH v2] Make TI EULA unpack work when /opt is writeable

ti-eula-unpack.inc: Use --prefix= to specify install path

ti-codecs-omap3530_1.00.01.bb:
ti-codecs-omap3530_1.01.00.bb:
ti-dspbios.inc:
ti-edma3lld.inc:
ti-xdctools.inc: Define TI_BIN_UNPK_WDEXT for proper unpacking

Note: only tested with these [recent] packages:
ti-cgt6x-1_6_1_9-r4
ti-cgt6x-1_6_1_14-r4
ti-codecs-omap3530-1_1_01_00-r2
ti-dspbios-5_41_04_18-r1
ti-edma3lld-01_11_00_03-r0
ti-xdctools-3_16_01_27-r2

Signed-off-by Gary Thomas <[email protected]>

---
 recipes/ti/ti-codecs-omap3530_1.00.01.bb |    1 +
 recipes/ti/ti-codecs-omap3530_1.01.00.bb |    1 +
 recipes/ti/ti-dspbios.inc                |    1 +
 recipes/ti/ti-edma3lld.inc               |    1 +
 recipes/ti/ti-eula-unpack.inc            |    5 ++++-
 recipes/ti/ti-xdctools.inc               |    1 +
 6 files changed, 9 insertions(+), 1 deletions(-)

recipes/ti/ti-codecs-omap3530_1.00.01.bb +1 -0

index d6d871b..37d2679 100644
@@ -26,6 +26,7 @@ SRC_URI = "http://software-dl.ti.com/dsps/dsps_public_sw/sdo_sb/targetcontent/dv
 
 BINFILE = "cs1omap3530_setuplinux_1_00_01-44.bin"
 TI_BIN_UNPK_CMDS = "Y:Y: qY:workdir"
+TI_BIN_UNPK_WDEXT="/dvsdk/dvsdk_3_00_02_44"
 
 DEPENDS = "ti-cgt6x ti-xdctools ti-dspbios ti-codec-engine ti-linuxutils"

recipes/ti/ti-codecs-omap3530_1.01.00.bb +1 -0

index 57bc7c4..e113883 100644
@@ -26,6 +26,7 @@ SRC_URI = "http://software-dl.ti.com/dsps/dsps_public_sw/sdo_tii/dvsdk/dvsdk_3_0
 
 BINFILE = "cs1omap3530_setupLinux_1_01_00-prebuilt-dvsdk3.01.00.10.bin"
 TI_BIN_UNPK_CMDS = "Y:Y: qY:workdir"
+TI_BIN_UNPK_WDEXT="/dvsdk/dvsdk_3_01_00_10"
 
 DEPENDS = "ti-cgt6x ti-xdctools ti-dspbios ti-codec-engine ti-linuxutils"

recipes/ti/ti-dspbios.inc +1 -0

index bd7bcc4..df71da1 100644
@@ -15,6 +15,7 @@ SRC_URI = "http://software-dl.ti.com/dsps/dsps_public_sw/sdo_sb/targetcontent/bi
 
 BINFILE="bios_setuplinux_${PV}.bin"
 TI_BIN_UNPK_CMDS="Y: qY:workdir:Y"
+TI_BIN_UNPK_WDEXT="/bios_${PV}"
 
 do_install() {
     install -d ${D}${BIOS_INSTALL_DIR_RECIPE}

recipes/ti/ti-edma3lld.inc +1 -0

index 94c0cff..344c357 100644
@@ -13,6 +13,7 @@ SRC_URI = "http://software-dl.ti.com/dsps/dsps_public_sw/sdo_sb/targetcontent/ed
 
 BINFILE="EDMA3_LLD_setuplinux_${PV}.bin"
 TI_BIN_UNPK_CMDS="Y:workdir"
+TI_BIN_UNPK_WDEXT="/edma3_lld_${PV}"
 
 do_install() {
     install -d ${D}${EDMA3_LLD_INSTALL_DIR_RECIPE}

recipes/ti/ti-eula-unpack.inc +4 -1

index a812a85..d215b6c 100644
@@ -30,6 +30,9 @@ python ti_bin_do_unpack() {
     workdir  = bb.data.getVar('WORKDIR', localdata)
     workdir  = bb.data.expand(workdir, localdata)
     os.chdir(workdir)
+    wdext  = bb.data.getVar('TI_BIN_UNPK_WDEXT', localdata)
+    wdext  = bb.data.expand(wdext, localdata)
+    wdext = workdir + '/' + wdext
 
     # Get unpack commands
     cmd_string = bb.data.getVar('TI_BIN_UNPK_CMDS', localdata)

@@ -39,7 +42,7 @@ python ti_bin_do_unpack() {
     os.chmod(binfile, 0755)
 
     # Run the InstallJammer binary and accept the EULA
-    filename = "HOME=%s ./%s --mode console" % (workdir, binfile)
+    filename = "HOME=%s ./%s --mode console --prefix %s" % (workdir, binfile, wdext)
     
     # Test executable by printing installer version or help screen (--version currently broken for some installers) 
     # - this is currently broken in some IJ installers - comment out for now

recipes/ti/ti-xdctools.inc +1 -0

index 4b451cf..c078e7c 100644
@@ -15,6 +15,7 @@ SRC_URI = "http://software-dl.ti.com/dsps/dsps_public_sw/sdo_sb/targetcontent/rt
 
 BINFILE="xdctools_setuplinux_${PV}.bin"
 TI_BIN_UNPK_CMDS="Y: qY:workdir:Y"
+TI_BIN_UNPK_WDEXT="/xdctools_${PV}"
 
 do_install() {
     install -d ${D}${XDC_INSTALL_DIR_RECIPE}

Signature · 1 line

1.6.2.5

Gary Thomas

author

2010-10-09 14:52 UTC #1 lore raw

Quoted 10 lines · Gary Thomas

This patch addresses the problem where the TI tools that need the EULA unpack into the wrong place if the user has write permission in /opt. The current behaviour of expecting the unpacker to fall back to the $HOME environment variable should not be relied on. The patch changes the unpack step to use the --prefix= option which will always work.
Version 2:
* Update patch to be against org.openembedded.dev branch
* Incorporate full description into patch

What about this patch? It's been nearly 6 weeks and no discussion or anything.

The problem persists in today's tree.

n.b. there was a discussion about automatic application of a patch if it goes more than two weeks without objection. What about that??

Signature · 4 lines

------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

Stefan Schmidt

2010-11-06 14:12 UTC #2 lore raw

Hello.

Quoted 8 lines · Gary Thomas

Quoted 7 lines · Gary Thomas
This patch addresses the problem where the TI tools that need the EULA unpack into the wrong place if the user has write permission in /opt. The current behaviour of expecting the unpacker to fall back to the $HOME environment variable should not be relied on. The patch changes the unpack step to use the --prefix= option which will always work.

Quoted 8 lines

Quoted 3 lines
Version 2:
* Update patch to be against org.openembedded.dev branch
* Incorporate full description into patch
What about this patch? It's been nearly 6 weeks and no discussion or anything.
The problem persists in today's tree.

Can we have a comment form the TI devs here?

regards
Stefan Schmidt

Koen Kooi

2010-11-07 09:06 UTC #3 lore raw

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Quoted 15 lines · Stefan Schmidt

Hello.
Quoted 12 lines · Gary Thomas
Quoted 10 lines · Gary Thomas
This patch addresses the problem where the TI tools that need the EULA unpack into the wrong place if the user has write permission in /opt. The current behaviour of expecting the unpacker to fall back to the $HOME environment variable should not be relied on. The patch changes the unpack step to use the --prefix= option which will always work.
Version 2:
* Update patch to be against org.openembedded.dev branch
* Incorporate full description into patch

Quoted 6 lines

Quoted 4 lines
What about this patch? It's been nearly 6 weeks and no discussion or anything.
The problem persists in today's tree.
Can we have a comment form the TI devs here?

I'm against automatic application of patches.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFM1mvQMkyGM64RGpERAjceAJ9CKgqOlknrKWrHs0uX45LbdWFlwwCeMvXy UCksWN+a/+jo4uNLQOu+rI0=
=SDiM
-----END PGP SIGNATURE-----

Stefan Schmidt

2010-11-07 11:26 UTC #4 lore raw

Hello.

Quoted 17 lines · Koen Kooi

Quoted 15 lines · Stefan Schmidt
Hello.
Quoted 12 lines · Gary Thomas
Quoted 10 lines · Gary Thomas
This patch addresses the problem where the TI tools that need the EULA unpack into the wrong place if the user has write permission in /opt. The current behaviour of expecting the unpacker to fall back to the $HOME environment variable should not be relied on. The patch changes the unpack step to use the --prefix= option which will always work.
Version 2:
* Update patch to be against org.openembedded.dev branch
* Incorporate full description into patch

Quoted 8 lines

Quoted 6 lines
Quoted 4 lines
What about this patch? It's been nearly 6 weeks and no discussion or anything.
The problem persists in today's tree.
Can we have a comment form the TI devs here?
I'm against automatic application of patches.

Interesting side note, but I was looking for a comment for the original issue. Writeable /opt and the problem the TI recipes have with this.

regards
Stefan Schmidt

Frans Meulenbroeks

2010-11-15 08:48 UTC #5 lore raw

In reply to Koen Kooi

2010/11/7 Koen Kooi <[email protected]>:

Quoted 19 lines

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Quoted 15 lines · Stefan Schmidt
Hello.
Quoted 12 lines · Gary Thomas
Quoted 10 lines · Gary Thomas
This patch addresses the problem where the TI tools that need the EULA unpack into the wrong place if the user has write permission in /opt. The current behaviour of expecting the unpacker to fall back to the $HOME environment variable should not be relied on. The patch changes the unpack step to use the --prefix= option which will always work.
Version 2:
* Update patch to be against org.openembedded.dev branch
* Incorporate full description into patch

Quoted 8 lines

Quoted 6 lines
Quoted 4 lines
What about this patch? It's been nearly 6 weeks and no discussion or anything.
The problem persists in today's tree.
Can we have a comment form the TI devs here?
I'm against automatic application of patches.

To me this clearly indicates the current problem with openembedded.

Apparenlty some people do not care on properly maintaining their recipes, addressing issues and patches raised by the community, reviewing and commenting on them.
But of course there is always time to grunt and whine and bash if others with less knowledge try to give attention to those patches.

Especially from a company like TI I had expected a somewhat more professional attitude wrt patches and improving their recipes.

I guess is that this is one of the areas where we as OE community loose when compared to yocto.

Frans

Maupin, Chase

2010-11-16 23:33 UTC #6 lore raw

In reply to Stefan Schmidt

Quoted 25 lines

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Stefan Schmidt
Sent: Saturday, November 06, 2010 9:11 AM
To: [email protected]
Subject: Re: [oe] [PATCH v2] TI tools - make EULA unpack work when /opt is writeable
Hello.
Quoted 12 lines · Gary Thomas
Quoted 10 lines · Gary Thomas
This patch addresses the problem where the TI tools that need the EULA unpack into the wrong place if the user has write permission in /opt. The current behaviour of expecting the unpacker to fall back to the $HOME environment variable should not be relied on. The patch changes the unpack step to use the --prefix= option which will always work.
Version 2:
* Update patch to be against org.openembedded.dev branch
* Incorporate full description into patch

Quoted 6 lines

Quoted 4 lines
What about this patch? It's been nearly 6 weeks and no discussion or anything.
The problem persists in today's tree.
Can we have a comment form the TI devs here?

This looks OK to me. Only comment I have is whether the '/' is needed in the following line with the TI_BIN_UNPK_WDEXT option already add a '/'. I guess while not technically necessary it would keep this code working if someone didn't define TI_BIN_UNPK_WDEXT with a leading '/'.

wdext = workdir + '/' + wdext

Sorry for the previous lack of response. Thank you for the patch and fixing this issue.

Quoted 8 lines

regards
Stefan Schmidt
_______________________________________________
Openembedded-devel mailing list
[email protected]
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel