lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAL_JsqLoLiq_cybL4Fk-DKkwKcu9DmJr9HcVfZEV2QvwV23kSg@mail.gmail.com>
Date:   Mon, 16 Oct 2017 15:45:52 -0500
From:   Rob Herring <robh+dt@...nel.org>
To:     Thomas Meyer <thomas@...3r.de>
Cc:     Frank Rowand <frowand.list@...il.com>,
        "devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: of: unittest fails on UML

On Fri, Oct 13, 2017 at 5:38 PM, Thomas Meyer <thomas@...3r.de> wrote:
> Hi,
>
> drivers/of/unittest.c fails on UML with a kernel crash.
> The crash is due to of_find_device_by_node() returning NULL:
>
> 841│         /* Test that a missing irq domain returns -EPROBE_DEFER */
> 842│         np = of_find_node_by_path("/testcase-data/testcase-device1");
> 843├>        pdev = of_find_device_by_node(np);
> 844│         unittest(pdev, "device 1 creation failed\n");
> 845│
> 846│         irq = platform_get_irq(pdev, 0);
> 847│         unittest(irq == -EPROBE_DEFER, "device deferred probe failed - %d\n", irq);
>
> in line 846 platform_get_irq is called with NULL.

The problem is IRQ_DOMAIN is not enabled.

>
> This helps to get through unittest.c:
>
> diff --git a/drivers/base/platform.c b/drivers/base/platform.c
> index d1bd99271066..a3b3d0ec8818 100644
> --- a/drivers/base/platform.c
> +++ b/drivers/base/platform.c
> @@ -86,6 +86,9 @@ EXPORT_SYMBOL_GPL(platform_get_resource);
>   */
>  int platform_get_irq(struct platform_device *dev, unsigned int num)
>  {
> +       if (!dev)
> +               return -ENXIO;
> +
>  #ifdef CONFIG_SPARC
>         /* sparc does not have irqs represented as IORESOURCE_IRQ resources */
>         if (!dev || num >= dev->archdata.num_irqs)
>
> log extract:
>
> [   24.850000] kobject: 'fragment@0' (000000006c818aa8): kobject_add_internal: parent: 'overlay4', set: 'devicetree'
> [   24.850000] ### dt-test ### start of unittest - you will see error messages
> [   24.850000] OF: /testcase-data/phandle-tests/consumer-a: could not get #phandle-cells-missing for /testcase-data/phandle-tests/provider1
> [   24.860000] OF: /testcase-data/phandle-tests/consumer-a: could not get #phandle-cells-missing for /testcase-data/phandle-tests/provider1
> [   24.860000] OF: /testcase-data/phandle-tests/consumer-a: could not find phandle
> [   24.860000] OF: /testcase-data/phandle-tests/consumer-a: could not find phandle
> [   24.860000] OF: /testcase-data/phandle-tests/consumer-a: arguments longer than property
> [   24.860000] OF: /testcase-data/phandle-tests/consumer-a: arguments longer than property
> [   24.860000] kobject: 'n1' (000000006cefca50): kobject_add_internal: parent: 'changeset', set: 'devicetree'
> [   24.860000] kobject: 'n2' (000000006cefde30): kobject_add_internal: parent: 'changeset', set: 'devicetree'
> [   24.860000] kobject: 'n21' (000000006cefcda0): kobject_add_internal: parent: 'n2', set: 'devicetree'
> [   24.860000] kobject: 'node-remove' (000000006c800880): kobject_add_internal: parent: 'changeset', set: 'devicetree'
> [   24.860000] kobject: 'n21' (000000006cefcda0): kobject_release, parent           (null) (delayed 400)
> [   24.860000] kobject: 'n2' (000000006cefde30): kobject_release, parent           (null) (delayed 100)
> [   24.860000] kobject: 'n1' (000000006cefca50): kobject_release, parent           (null) (delayed 300)
> [   24.860000] ### dt-test ### FAIL of_unittest_platform_populate():844 device 1 creation failed
> [   24.860000] ### dt-test ### FAIL of_unittest_platform_populate():847 device deferred probe failed - -6
> [   24.860000] ### dt-test ### FAIL of_unittest_platform_populate():852 device 2 creation failed
> [   24.860000] Registering platform device 'unittest-bus.0'. Parent at platform
> [   24.860000] device: 'unittest-bus.0': device_add
> [   24.860000] kobject: 'unittest-bus.0' (000000006cf11568): kobject_add_internal: parent: 'platform', set: 'devices'
> [   24.860000] bus: 'platform': add device unittest-bus.0
> [   24.860000] kobject: 'unittest-bus.0' (000000006cf11568): kobject_uevent_env
> [   24.860000] kobject: 'unittest-bus.0' (000000006cf11568): fill_kobj_path: path = '/devices/platform/unittest-bus.0'
> [   24.860000] ### dt-test ### FAIL of_unittest_platform_populate():882 Could not create device for node 'dev'
> [   24.860000] ### dt-test ### FAIL of_unittest_platform_populate():882 Could not create device for node 'dev'
> [   24.860000] bus: 'platform': remove device unittest-bus.0
> [   24.860000] kobject: 'unittest-bus.0' (000000006cf11568): kobject_uevent_env
> [   24.860000] kobject: 'unittest-bus.0' (000000006cf11568): fill_kobj_path: path = '/devices/platform/unittest-bus.0'
> [   24.860000] kobject: 'unittest-bus.0' (000000006cf11568): kobject_release, parent           (null) (delayed 100)
> [   24.860000] bus: 'platform': add driver unittest
> [   24.860000] kobject: 'unittest' (000000006c86bba8): kobject_add_internal: parent: 'drivers', set: 'drivers'
> [   24.870000] kobject: 'unittest' (000000006c86bba8): kobject_uevent_env
> [   24.870000] kobject: 'unittest' (000000006c86bba8): fill_kobj_path: path = '/bus/platform/drivers/unittest'
> [   24.870000] ### dt-test ### FAIL of_unittest_overlay():1944 could not populate bus @ "/testcase-data/overlay-node/test-bus"
> [   24.870000] ### dt-test ### FAIL of_unittest_overlay_high_level():2201 overlay_base_root not initialized
> [   24.870000] ### dt-test ### end of unittest - 163 passed, 7 failed
>
> so any idea what is going on here?
> any help is appreciated.

I got things working a while back with this (whitespace damaged)
patch. I'm not all that happy with it. Perhaps the unittest should
select IRQ_DOMAIN instead. Really, I think HAS_IOMEM should die. It is
true for practically every arch (all but tile, score, and um) and
really only controls enabling some drivers or not. It selects
GENERIC_IO which controls MTD visibility and nothing else.

commit 9833012b9290292066127890df07eaccae4ee6e3
Author: Rob Herring <robh@...nel.org>
Date:   Tue Jul 25 14:46:56 2017 -0500

    UML fixes for DT unittests

    Signed-off-by: Rob Herring <robh@...nel.org>

diff --git a/arch/um/Kconfig.common b/arch/um/Kconfig.common
index 85f6dd204ab6..c3ff9a3b4503 100644
--- a/arch/um/Kconfig.common
+++ b/arch/um/Kconfig.common
@@ -12,6 +12,7 @@ config UML
        select GENERIC_IO
        select GENERIC_CLOCKEVENTS
        select HAVE_GCC_PLUGINS
+       select IRQ_DOMAIN
        select TTY # Needed for line.c

 config MMU
diff --git a/drivers/of/Kconfig b/drivers/of/Kconfig
index ba7b034b2b91..a688478d763e 100644
--- a/drivers/of/Kconfig
+++ b/drivers/of/Kconfig
@@ -57,7 +57,7 @@ config OF_DYNAMIC

 config OF_ADDRESS
        def_bool y
-       depends on !SPARC && HAS_IOMEM
+       depends on !SPARC && (HAS_IOMEM || UML)
        select OF_ADDRESS_PCI if PCI

 config OF_ADDRESS_PCI

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ