| Message ID | 118eb454-703f-9de9-6fc4-5326de6592ca@broadcom.com (mailing list archive) |
|---|---|
| State | Not Applicable |
| Headers | show |
On 10-1-2017 12:16, Arend Van Spriel wrote: > With 4.10-rc1 the devcoredump.patch in patches/backport-adjustment no > longer applies due to following commit: > > commit f76d25275c314defb684fdd692239507001774bc > Author: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > Date: Mon Nov 28 16:41:58 2016 +0100 > > driver core: devcoredump: convert to use class_groups > > Convert devcoredump to use class_groups instead of class_attrs as that's > the correct way to handle lists of class attribute files. > > Acked-by: Johannes Berg <johannes@sipsolutions.net> > Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> > > So I am reworking the patch file (attached). When I run gentree I get > following: > > Traceback (most recent call last): > File "./gentree.py", line 1091, in <module> > ret = _main() > File "./gentree.py", line 724, in _main > logwrite=logwrite) > File "./gentree.py", line 906, in process > apply_patches(args, "backport", source_dir, 'patches', > bpid.target_dir, logwrite) > File "./gentree.py", line 514, in apply_patches > raise Exception('No patch content found in %s' % print_name) > Exception: No patch content found in backport-adjustments/devcoredump.patch > > However, go into the target dir and applying patch works fine: > > $ patch -p1 < ~-/patches/backport-adjustments/devcoredump.patch > patching file compat/drivers-base-devcoredump.c > Hunk #7 succeeded at 181 with fuzz 1 (offset 2 lines). > Hunk #8 succeeded at 248 (offset 2 lines). > Hunk #9 succeeded at 339 (offset 2 lines). > Hunk #10 succeeded at 386 (offset 2 lines). > patching file include/linux/backport-devcoredump.h > > So I am a bit puzzled as to what is going on so currently debugging > lib/patch.py. If anyone has a good hint feel free to let me know. Turns out that at line 70 of the patch file there is no single space before the tab, which lib/patch.py marks as an error. With that fixed the fun really starts as I get following code: static CLASS_ATTR_RW(disabled); static struct attribute *devcd_class_attrs[] = { &class_attr_disabled.attr, NULL, }; #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) ATTRIBUTE_GROUPS(devcd_class); #else #define BP_ATTR_GRP_STRUCT device_attribute ATTRIBUTE_GROUPS_BACKPORT(devcd_class); #endif static struct class devcd_class = { .name = "devcoredump", .owner = THIS_MODULE, .dev_release = devcd_dev_release, #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) .dev_groups = devcd_dev_groups, #else .dev_attrs = devcd_class_dev_attrs, #endif #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(4,10,0) .class_attrs = devcd_class_attrs, #else .class_groups = devcd_class_groups, #endif }; This is because the new devcoredump.c in 4.10-rc1 now triggers a Coccinelle script to kick in, ie. patches/collateral-evolutions/generic/0001-group-attr/0001-group_attr_class.cocci. It seems to kick in because of the ATTRIBUTE_GROUPS(devcd_class) macro. Regards, Arend -- To unsubscribe from this list: send the line "unsubscribe backports" in
Hey Arend, On Tue, 2017-01-10 at 21:39 +0100, Arend Van Spriel wrote: > Turns out that at line 70 of the patch file there is no single space > before the tab, which lib/patch.py marks as an error. With that fixed > the fun really starts as I get following code: > > static CLASS_ATTR_RW(disabled); > > static struct attribute *devcd_class_attrs[] = { > &class_attr_disabled.attr, > NULL, > }; > #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) > ATTRIBUTE_GROUPS(devcd_class); > #else > #define BP_ATTR_GRP_STRUCT device_attribute > ATTRIBUTE_GROUPS_BACKPORT(devcd_class); > #endif > > static struct class devcd_class = { > .name = "devcoredump", > .owner = THIS_MODULE, > .dev_release = devcd_dev_release, > #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) > #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) > .dev_groups = devcd_dev_groups, > #else > .dev_attrs = devcd_class_dev_attrs, > #endif > #endif > #if LINUX_VERSION_CODE < KERNEL_VERSION(4,10,0) > .class_attrs = devcd_class_attrs, > #else > .class_groups = devcd_class_groups, > #endif > }; > > This is because the new devcoredump.c in 4.10-rc1 now triggers a > Coccinelle script to kick in, ie. > patches/collateral-evolutions/generic/0001-group-attr/0001-group_attr_class.cocci. > It seems to kick in because of the ATTRIBUTE_GROUPS(devcd_class) macro. Did you ever get this to work? I'm getting this warnings: home/luca/iwlwifi/stack-dev/compat/drivers-base-devcoredump.c:197:3: warning: initialization from incompatible pointer type [enabled by default] .class_attrs = devcd_class_attrs, ^ /home/luca/iwlwifi/stack-dev/compat/drivers-base-devcoredump.c:197:3: warning: (near initialization for ‘devcd_class.class_attrs’) [enabled by default] In file included from /home/luca/iwlwifi/stack-dev/backport-include/linux/sysfs.h:3:0, from include/linux/kobject.h:21, from include/linux/module.h:17, from /home/luca/iwlwifi/stack-dev/backport-include/linux/module.h:3, from /home/luca/iwlwifi/stack-dev/compat/drivers-base-devcoredump.c:27: /home/luca/iwlwifi/stack-dev/compat/drivers-base-devcoredump.c:179:18: warning: ‘devcd_class_groups’ defined but not used [-Wunused-variable] ATTRIBUTE_GROUPS(devcd_class); ^ include/linux/sysfs.h:131:38: note: in definition of macro ‘__ATTRIBUTE_GROUPS’ static const struct attribute_group *_name##_groups[] = { \ ^ /home/luca/iwlwifi/stack-dev/compat/drivers-base-devcoredump.c:179:1: note: in expansion of macro ‘ATTRIBUTE_GROUPS’ ATTRIBUTE_GROUPS(devcd_class); ^ Any idea how to fix it? -- Cheers, Luca. -- To unsubscribe from this list: send the line "unsubscribe backports" in
On 28-1-2017 11:50, Luca Coelho wrote: > Hey Arend, > > On Tue, 2017-01-10 at 21:39 +0100, Arend Van Spriel wrote: >> Turns out that at line 70 of the patch file there is no single space >> before the tab, which lib/patch.py marks as an error. With that fixed >> the fun really starts as I get following code: >> >> static CLASS_ATTR_RW(disabled); >> >> static struct attribute *devcd_class_attrs[] = { >> &class_attr_disabled.attr, >> NULL, >> }; >> #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) >> ATTRIBUTE_GROUPS(devcd_class); >> #else >> #define BP_ATTR_GRP_STRUCT device_attribute >> ATTRIBUTE_GROUPS_BACKPORT(devcd_class); >> #endif >> >> static struct class devcd_class = { >> .name = "devcoredump", >> .owner = THIS_MODULE, >> .dev_release = devcd_dev_release, >> #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) >> #if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) >> .dev_groups = devcd_dev_groups, >> #else >> .dev_attrs = devcd_class_dev_attrs, >> #endif >> #endif >> #if LINUX_VERSION_CODE < KERNEL_VERSION(4,10,0) >> .class_attrs = devcd_class_attrs, >> #else >> .class_groups = devcd_class_groups, >> #endif >> }; >> >> This is because the new devcoredump.c in 4.10-rc1 now triggers a >> Coccinelle script to kick in, ie. >> patches/collateral-evolutions/generic/0001-group-attr/0001-group_attr_class.cocci. >> It seems to kick in because of the ATTRIBUTE_GROUPS(devcd_class) macro. > > Did you ever get this to work? I'm getting this warnings: > > home/luca/iwlwifi/stack-dev/compat/drivers-base-devcoredump.c:197:3: warning: initialization from incompatible pointer type [enabled by default] > .class_attrs = devcd_class_attrs, > ^ > /home/luca/iwlwifi/stack-dev/compat/drivers-base-devcoredump.c:197:3: warning: (near initialization for ‘devcd_class.class_attrs’) [enabled by default] > In file included from /home/luca/iwlwifi/stack-dev/backport-include/linux/sysfs.h:3:0, > from include/linux/kobject.h:21, > from include/linux/module.h:17, > from /home/luca/iwlwifi/stack-dev/backport-include/linux/module.h:3, > from /home/luca/iwlwifi/stack-dev/compat/drivers-base-devcoredump.c:27: > /home/luca/iwlwifi/stack-dev/compat/drivers-base-devcoredump.c:179:18: warning: ‘devcd_class_groups’ defined but not used [-Wunused-variable] > ATTRIBUTE_GROUPS(devcd_class); > ^ > include/linux/sysfs.h:131:38: note: in definition of macro ‘__ATTRIBUTE_GROUPS’ > static const struct attribute_group *_name##_groups[] = { \ > ^ > /home/luca/iwlwifi/stack-dev/compat/drivers-base-devcoredump.c:179:1: note: in expansion of macro ‘ATTRIBUTE_GROUPS’ > ATTRIBUTE_GROUPS(devcd_class); > ^ > > Any idea how to fix it? Yeah. This is what I hit after fixing the devcoredump.patch. Because of the recent change in drivers/base/devcoredump.c I think 0001-group_attr_class.cocci needs to be quite a bit smarter. My SmPL knowledge is still somewhat limited so I did not get it working properly. Maybe posting the problem (in simplified form) on the Cocci mailing list will get us a solution. Regards, Arend -- To unsubscribe from this list: send the line "unsubscribe backports" in
diff --git a/compat/drivers-base-devcoredump.c b/compat/drivers-base-devcoredump.c index 240374f..8bb376b 100644 --- a/compat/drivers-base-devcoredump.c +++ b/compat/drivers-base-devcoredump.c @@ -31,6 +31,7 @@ #include <linux/slab.h> #include <linux/fs.h> #include <linux/workqueue.h> +#include "backports.h" static struct class devcd_class; @@ -40,6 +41,10 @@ static bool devcd_disabled; /* if data isn't read by userspace after 5 minutes then delete it */ #define DEVCD_TIMEOUT (HZ * 60 * 5) +#if LINUX_VERSION_CODE < KERNEL_VERSION(3,11,0) +static struct bin_attribute devcd_attr_data; +#endif + struct devcd_entry { struct device devcd_dev; void *data; @@ -69,8 +74,7 @@ static void devcd_dev_release(struct device *dev) * a struct device to know when it goes away? */ if (devcd->failing_dev->kobj.sd) - sysfs_delete_link(&devcd->failing_dev->kobj, &dev->kobj, - "devcoredump"); + sysfs_remove_link(&devcd->failing_dev->kobj, "devcoredump"); put_device(devcd->failing_dev); kfree(devcd); @@ -82,6 +86,9 @@ static void devcd_del(struct work_struct *wk) devcd = container_of(wk, struct devcd_entry, del_wk.work); +#if LINUX_VERSION_CODE < KERNEL_VERSION(3,11,0) + device_remove_bin_file(&devcd->devcd_dev, &devcd_attr_data); +#endif device_del(&devcd->devcd_dev); put_device(&devcd->devcd_dev); } @@ -115,6 +122,7 @@ static struct bin_attribute devcd_attr_data = { .write = devcd_data_write, }; +#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) static struct bin_attribute *devcd_dev_bin_attrs[] = { &devcd_attr_data, NULL, }; @@ -126,6 +134,7 @@ static const struct attribute_group devcd_dev_group = { static const struct attribute_group *devcd_dev_groups[] = { &devcd_dev_group, NULL, }; +#endif /* LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) */ static int devcd_free(struct device *dev, void *data) { @@ -170,7 +179,13 @@ static struct class devcd_class = { .name = "devcoredump", .owner = THIS_MODULE, .dev_release = devcd_dev_release, +#if LINUX_VERSION_CODE >= KERNEL_VERSION(3,11,0) .dev_groups = devcd_dev_groups, +#endif +#if LINUX_VERSION_CODE < KERNEL_VERSION(4,10,0) + .class_attrs = devcd_class_attrs, +#else .class_groups = devcd_class_groups, +#endif }; @@ -231,6 +242,14 @@ static void devcd_free_sgtable(void *data) _devcd_free_sgtable(data); } +#if LINUX_VERSION_CODE < KERNEL_VERSION(3,11,0) +size_t sg_pcopy_to_buffer(struct scatterlist *sgl, unsigned int nents, + void *buf, size_t buflen, off_t skip) +{ + return 0; +} +#endif + /** * devcd_read_from_table - copy data from sg_table to a given buffer * and return the number of bytes read @@ -314,6 +333,11 @@ void dev_coredumpm(struct device *dev, struct module *owner, if (device_add(&devcd->devcd_dev)) goto put_device; +#if LINUX_VERSION_CODE < KERNEL_VERSION(3,11,0) + if (device_create_bin_file(&devcd->devcd_dev, &devcd_attr_data)) + goto put_device; +#endif + if (sysfs_create_link(&devcd->devcd_dev.kobj, &dev->kobj, "failing_device")) /* nothing - symlink will be missing */; @@ -356,15 +380,13 @@ void dev_coredumpsg(struct device *dev, struct scatterlist *table, } EXPORT_SYMBOL_GPL(dev_coredumpsg); -static int __init devcoredump_init(void) +int __init devcoredump_init(void) { return class_register(&devcd_class); } -__initcall(devcoredump_init); -static void __exit devcoredump_exit(void) +void __exit devcoredump_exit(void) { class_for_each_device(&devcd_class, NULL, NULL, devcd_free); class_unregister(&devcd_class); } -__exitcall(devcoredump_exit); --- a/include/linux/backport-devcoredump.h +++ b/include/linux/backport-devcoredump.h @@ -66,7 +66,7 @@ static inline void _devcd_free_sgtable(struct scatterlist *table) } -#ifdef CONFIG_DEV_COREDUMP +#ifdef CPTCFG_BPAUTO_WANT_DEV_COREDUMP void dev_coredumpv(struct device *dev, void *data, size_t datalen, gfp_t gfp); @@ -100,6 +100,6 @@ static inline void dev_coredumpsg(struct device *dev, struct scatterlist *table, { _devcd_free_sgtable(table); } -#endif /* CONFIG_DEV_COREDUMP */ +#endif /* CPTCFG_BPAUTO_WANT_DEV_COREDUMP */ #endif /* __DEVCOREDUMP_H */
With 4.10-rc1 the devcoredump.patch in patches/backport-adjustment no longer applies due to following commit: commit f76d25275c314defb684fdd692239507001774bc Author: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Date: Mon Nov 28 16:41:58 2016 +0100 driver core: devcoredump: convert to use class_groups Convert devcoredump to use class_groups instead of class_attrs as that's the correct way to handle lists of class attribute files. Acked-by: Johannes Berg <johannes@sipsolutions.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> So I am reworking the patch file (attached). When I run gentree I get following: Traceback (most recent call last): File "./gentree.py", line 1091, in <module> ret = _main() File "./gentree.py", line 724, in _main logwrite=logwrite) File "./gentree.py", line 906, in process apply_patches(args, "backport", source_dir, 'patches', bpid.target_dir, logwrite) File "./gentree.py", line 514, in apply_patches raise Exception('No patch content found in %s' % print_name) Exception: No patch content found in backport-adjustments/devcoredump.patch However, go into the target dir and applying patch works fine: $ patch -p1 < ~-/patches/backport-adjustments/devcoredump.patch patching file compat/drivers-base-devcoredump.c Hunk #7 succeeded at 181 with fuzz 1 (offset 2 lines). Hunk #8 succeeded at 248 (offset 2 lines). Hunk #9 succeeded at 339 (offset 2 lines). Hunk #10 succeeded at 386 (offset 2 lines). patching file include/linux/backport-devcoredump.h So I am a bit puzzled as to what is going on so currently debugging lib/patch.py. If anyone has a good hint feel free to let me know. Regards, Arend