[meta-raspberrypi][PATCH] sdcard_image-rpi: Prevent taskhash mismatch

[meta-raspberrypi][PATCH] sdcard_image-rpi: Prevent taskhash mismatch

[Paul Barker]

As recently discussed on the mailing list, bitbake now issues an error when the
task hash computed by the bitbake master differs from the task hash computed by
the bitbake worker. This usually happens when the task hash depends on the date
and time for some reason.

This affects IMAGE_CMD_rpi-sdimg and is fixed by excluding the dependency on the
current date and time from the command.

Signed-off-by: Paul Barker <paul@paulbarker.me.uk>
---
 classes/sdcard_image-rpi.bbclass | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/classes/sdcard_image-rpi.bbclass b/classes/sdcard_image-rpi.bbclass
index 3b4f13f..a8d83a3 100644
--- a/classes/sdcard_image-rpi.bbclass
+++ b/classes/sdcard_image-rpi.bbclass
@@ -72,7 +72,7 @@ SDIMG = "${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.rpi-sdimg"
 FATPAYLOAD ?= ""
 
 IMAGEDATESTAMP = "${@time.strftime('%Y.%m.%d',time.gmtime())}"
-IMAGE_CMD_rpi-sdimg[vardepsexclude] = "IMAGEDATESTAMP"
+IMAGE_CMD_rpi-sdimg[vardepsexclude] = "DATETIME"
 
 IMAGE_CMD_rpi-sdimg () {
 
-- 
2.1.4

Re: [meta-raspberrypi][PATCH] sdcard_image-rpi: Prevent taskhash mismatch

[Andrei Gherzan]

[-- Attachment #1: Type: text/plain, Size: 1299 bytes --]

Hi Paul,

On Wed, Jun 08, 2016 at 02:02:29PM +0100, Paul Barker wrote:
> As recently discussed on the mailing list, bitbake now issues an error when the
> task hash computed by the bitbake master differs from the task hash computed by
> the bitbake worker. This usually happens when the task hash depends on the date
> and time for some reason.
>
> This affects IMAGE_CMD_rpi-sdimg and is fixed by excluding the dependency on the
> current date and time from the command.
>
> Signed-off-by: Paul Barker <paul@paulbarker.me.uk>
> ---
>  classes/sdcard_image-rpi.bbclass | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/classes/sdcard_image-rpi.bbclass b/classes/sdcard_image-rpi.bbclass
> index 3b4f13f..a8d83a3 100644
> --- a/classes/sdcard_image-rpi.bbclass
> +++ b/classes/sdcard_image-rpi.bbclass
> @@ -72,7 +72,7 @@ SDIMG = "${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.rpi-sdimg"
>  FATPAYLOAD ?= ""
>
>  IMAGEDATESTAMP = "${@time.strftime('%Y.%m.%d',time.gmtime())}"
> -IMAGE_CMD_rpi-sdimg[vardepsexclude] = "IMAGEDATESTAMP"
> +IMAGE_CMD_rpi-sdimg[vardepsexclude] = "DATETIME"
>
>  IMAGE_CMD_rpi-sdimg () {

I can't seem to replicate this issue with currrent poky master. Any idea
why? Maybe this got fixed in poky?



--
Andrei Gherzan

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [meta-raspberrypi][PATCH] sdcard_image-rpi: Prevent taskhash mismatch

[Paul Barker]

On Tue, 14 Jun 2016 19:30:25 +0100
Andrei Gherzan <andrei@gherzan.ro> wrote:

> Hi Paul,
> 
> On Wed, Jun 08, 2016 at 02:02:29PM +0100, Paul Barker wrote:
> > As recently discussed on the mailing list, bitbake now issues an
> > error when the task hash computed by the bitbake master differs
> > from the task hash computed by the bitbake worker. This usually
> > happens when the task hash depends on the date and time for some
> > reason.
> >
> > This affects IMAGE_CMD_rpi-sdimg and is fixed by excluding the
> > dependency on the current date and time from the command.
> >
> > Signed-off-by: Paul Barker <paul@paulbarker.me.uk>
> > ---
> >  classes/sdcard_image-rpi.bbclass | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/classes/sdcard_image-rpi.bbclass
> > b/classes/sdcard_image-rpi.bbclass index 3b4f13f..a8d83a3 100644
> > --- a/classes/sdcard_image-rpi.bbclass
> > +++ b/classes/sdcard_image-rpi.bbclass
> > @@ -72,7 +72,7 @@ SDIMG =
> > "${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.rpi-sdimg" FATPAYLOAD ?=
> > ""
> >
> >  IMAGEDATESTAMP = "${@time.strftime('%Y.%m.%d',time.gmtime())}"
> > -IMAGE_CMD_rpi-sdimg[vardepsexclude] = "IMAGEDATESTAMP"
> > +IMAGE_CMD_rpi-sdimg[vardepsexclude] = "DATETIME"
> >
> >  IMAGE_CMD_rpi-sdimg () {
> 
> I can't seem to replicate this issue with currrent poky master. Any
> idea why? Maybe this got fixed in poky?
> 

I must admit it's an issue I don't 100% understand. I'll try again
without this patch and see what happens.

Thanks,
Paul Barker

Re: [meta-raspberrypi][PATCH] sdcard_image-rpi: Prevent taskhash mismatch

[Paul Barker]

On Wed, 15 Jun 2016 12:05:53 +0100
Paul Barker <paul@paulbarker.me.uk> wrote:

> On Tue, 14 Jun 2016 19:30:25 +0100
> Andrei Gherzan <andrei@gherzan.ro> wrote:
> 
> > Hi Paul,
> > 
> > On Wed, Jun 08, 2016 at 02:02:29PM +0100, Paul Barker wrote:
> > > As recently discussed on the mailing list, bitbake now issues an
> > > error when the task hash computed by the bitbake master differs
> > > from the task hash computed by the bitbake worker. This usually
> > > happens when the task hash depends on the date and time for some
> > > reason.
> > >
> > > This affects IMAGE_CMD_rpi-sdimg and is fixed by excluding the
> > > dependency on the current date and time from the command.
> > >
> > > Signed-off-by: Paul Barker <paul@paulbarker.me.uk>
> > > ---
> > >  classes/sdcard_image-rpi.bbclass | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/classes/sdcard_image-rpi.bbclass
> > > b/classes/sdcard_image-rpi.bbclass index 3b4f13f..a8d83a3 100644
> > > --- a/classes/sdcard_image-rpi.bbclass
> > > +++ b/classes/sdcard_image-rpi.bbclass
> > > @@ -72,7 +72,7 @@ SDIMG =
> > > "${DEPLOY_DIR_IMAGE}/${IMAGE_NAME}.rootfs.rpi-sdimg" FATPAYLOAD ?=
> > > ""
> > >
> > >  IMAGEDATESTAMP = "${@time.strftime('%Y.%m.%d',time.gmtime())}"
> > > -IMAGE_CMD_rpi-sdimg[vardepsexclude] = "IMAGEDATESTAMP"
> > > +IMAGE_CMD_rpi-sdimg[vardepsexclude] = "DATETIME"
> > >
> > >  IMAGE_CMD_rpi-sdimg () {
> > 
> > I can't seem to replicate this issue with currrent poky master. Any
> > idea why? Maybe this got fixed in poky?
> > 
> 
> I must admit it's an issue I don't 100% understand. I'll try again
> without this patch and see what happens.

Weird... there's now no error even without this patch. I can't see
anything in bitbake/oe-core that I understand as a fix for this but it
may have been fixed as a side effect of something else. I'll keep an
eye out and see if the issue re-occurs in the future.

Thanks,
Paul Barker