Re: [PATCH 1/1] linux-yocto/3.4: v3.4.10 and uprobes/kprobes configuration updates

[Bruce Ashfield <bruce.ashfield@windriver.com>]

On 12-09-11 08:37 AM, Martin Jansa wrote:
> On Tue, Sep 11, 2012 at 08:31:42AM -0400, Bruce Ashfield wrote:
>> On 12-09-11 02:59 AM, Martin Jansa wrote:
>>> On Mon, Sep 10, 2012 at 8:11 PM, Bruce Ashfield
>>> <bruce.ashfield@windriver.com>   wrote:
>>>> Updating to 3.4.10 which has been soaking for a bit now, as well
>>>> as picking up the following meta commits from Tom Z:
>>>>
>>>>     a82db2f meta: have systemtap use kprobes and uprobes feature
>>>>     d5d5b80 meta: add kprobes support to ktypes/standard
>>>>     b32d373 meta: add kprobes feature
>>>>     d40ed99 meta: have uprobe feature use uprobe.cfg
>>>>     a69d1db meta: add uprobe.cfg
>>>>
>>>> Signed-off-by: Bruce Ashfield<bruce.ashfield@windriver.com>
>>>> ---
>>>>    meta/recipes-kernel/linux/linux-yocto-rt_3.4.bb |    8 ++++----
>>>>    meta/recipes-kernel/linux/linux-yocto_3.4.bb    |   16 ++++++++--------
>>>>    2 files changed, 12 insertions(+), 12 deletions(-)
>>>>
>>>> diff --git a/meta/recipes-kernel/linux/linux-yocto-rt_3.4.bb b/meta/recipes-kernel/linux/linux-yocto-rt_3.4.bb
>>>> index b2620ea..3b36378 100644
>>>> --- a/meta/recipes-kernel/linux/linux-yocto-rt_3.4.bb
>>>> +++ b/meta/recipes-kernel/linux/linux-yocto-rt_3.4.bb
>>>> @@ -3,14 +3,14 @@ require recipes-kernel/linux/linux-yocto.inc
>>>>    KBRANCH = "standard/preempt-rt/base"
>>>>    KBRANCH_qemuppc = "standard/preempt-rt/qemuppc"
>>>>
>>>> -LINUX_VERSION ?= "3.4.9"
>>>> +LINUX_VERSION ?= "3.4.10"
>>>>    LINUX_KERNEL_TYPE = "preempt-rt"
>>>>
>>>>    KMETA = "meta"
>>>>
>>>> -SRCREV_machine ?= "9032b1e9daf5b4396f939981c3be95f67802d18c"
>>>> -SRCREV_machine_qemuppc ?= "08ce190232f89303772b6591ca7daaf2820eb74e"
>>>> -SRCREV_meta ?= "463299bc2e533e1bd38b0053ae7b210980f269c3"
>>>> +SRCREV_machine ?= "a35693b1287c0e50cdca33a1b95af0ff48b43cd0"
>>>> +SRCREV_machine_qemuppc ?= "85a1190530cb5749f5f831670976b163438dc301"
>>>> +SRCREV_meta ?= "d9d5fc63d8b38705036e946ea77d971d95de11ad"
>>>>
>>>>    PR = "${INC_PR}.0"
>>>>    PV = "${LINUX_VERSION}+git${SRCPV}"
>>>> diff --git a/meta/recipes-kernel/linux/linux-yocto_3.4.bb b/meta/recipes-kernel/linux/linux-yocto_3.4.bb
>>>> index 06bcb9a..7258cba 100644
>>>> --- a/meta/recipes-kernel/linux/linux-yocto_3.4.bb
>>>> +++ b/meta/recipes-kernel/linux/linux-yocto_3.4.bb
>>>> @@ -3,17 +3,17 @@ require recipes-kernel/linux/linux-yocto.inc
>>>>    KBRANCH_DEFAULT = "standard/base"
>>>>    KBRANCH = "${KBRANCH_DEFAULT}"
>>>>
>>>> -SRCREV_machine_qemuarm ?= "84d8ee32265eea5d60f57a2f70bd3b9a0fb9213d"
>>>> -SRCREV_machine_qemumips  ?= "ba0e336d4527080233c3c410989d4f351529ee4e"
>>>> -SRCREV_machine_qemuppc ?= "e82b8a111430e3820b11f507863c4b8e8734ed8e"
>>>> -SRCREV_machine_qemux86 ?= "0985844fa6235422c67ef269952fa4e765f252f9"
>>>> -SRCREV_machine_qemux86-64 ?= "0985844fa6235422c67ef269952fa4e765f252f9"
>>>> -SRCREV_machine ?= "0985844fa6235422c67ef269952fa4e765f252f9"
>>>> -SRCREV_meta ?= "463299bc2e533e1bd38b0053ae7b210980f269c3"
>>>> +SRCREV_machine_qemuarm ?= "b15e7b1e9b58b9863bd87778775f86cd8d8880ea"
>>>> +SRCREV_machine_qemumips  ?= "8d5b98f263b5119af2dc30223f311be17173bab9"
>>>> +SRCREV_machine_qemuppc ?= "b9a720ca38d298ed457f37d099c85771f9164b19"
>>>> +SRCREV_machine_qemux86 ?= "46d8c757b3be1953f30d6745505d24436e2d6844"
>>>> +SRCREV_machine_qemux86-64 ?= "46d8c757b3be1953f30d6745505d24436e2d6844"
>>>> +SRCREV_machine ?= "46d8c757b3be1953f30d6745505d24436e2d6844"
>>>> +SRCREV_meta ?= "a82db2f0fc3ceebf3cb47e9dd05e4856ff9966ab"
>>>
>>> Doesn't different SRCREV_machine for each MACHINE cause LOCALCOUNT in
>>> SRCPV incrementing on each MACHINE switch?
>>>
>>> They are stored under same key:
>>> sqlite>   select * from BB_URI_LOCALCOUNT where key like '%linux-yocto-3.4%';
>>> git:git.yoctoproject.org.linux-yocto-3.4.gitstandard/base-linux-yocto_rev|84d8ee32265eea5d60f57a2f70bd3b9a0fb9213d
>>> git:git.yoctoproject.org.linux-yocto-3.4.gitstandard/base-linux-yocto_count|16
>>> git:git.yoctoproject.org.linux-yocto-3.4.gitmeta-linux-yocto_rev|463299bc2e533e1bd38b0053ae7b210980f269c3
>>> git:git.yoctoproject.org.linux-yocto-3.4.gitmeta-linux-yocto_count|9
>>> git:git.yoctoproject.org.linux-yocto-3.4.gitstandard/common-pc-64/base-linux-yocto_rev|19f7e43b54aef08d58135ed2a897d77b624b320a
>>> git:git.yoctoproject.org.linux-yocto-3.4.gitstandard/common-pc-64/base-linux-yocto_count|7
>>> git:git.yoctoproject.org.linux-yocto-3.4.gitstandard/preempt-rt/base-linux-yocto-rt_rev|9032b1e9daf5b4396f939981c3be95f67802d18c
>>> git:git.yoctoproject.org.linux-yocto-3.4.gitstandard/preempt-rt/base-linux-yocto-rt_count|10
>>> git:git.yoctoproject.org.linux-yocto-3.4.gitmeta-linux-yocto-rt_rev|463299bc2e533e1bd38b0053ae7b210980f269c3
>>> git:git.yoctoproject.org.linux-yocto-3.4.gitmeta-linux-yocto-rt_count|8
>>>
>>> So I guess if you build linux-yocto-3.4 for qemuarm, qemux86-64, then
>>> switch back to qemuarm you'll see linux-yocto built again, same
>>> source, but different SRCPV (LOCALCOUNT).
>>
>> That does look to be the case, and it matches what I've observed
>> over time. I'm not sure of an alternative at the moment, since the
>> fetcher is such a cranky beast with respect to fetching changes to
>> the right machine branches.
>
> Why not remove it from SRCREV_FORMAT, keep only meta SRCPV and just bump
> PR when SRCREV of some machine kbranch is changed?

I like the sound of it, but as far as I know, wouldn't that fix the
package revision, but cause the fetcher problems ? I've added Richard
to the cc, since I'm not sure what the current status of the fetcher
is in this regard.

If I don't bump the SRCREV for the machines, and just bump the PR,
there's nothing in place to trigger the fetcher when only machine
changes have been made to one of the branches, and conversely there's
no explicit value set to inhibit a machine's SRCREV when you don't
want the end of the branch to be built (I always build the end, but
other layers inhibit SRCREV frequently to throttle their updates).

If there's a better way that keeps the rebuilds to a minimum, but
doesn't break the fetcher and those other use cases .. I'm more than
happy to switch to it :)

Bruce

>
> Current SRCREV_FORMAT doesn't show which branch it was so it doesn't
> make it much worse to find what sources were used to build.
>
> Cheers,
>