summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2017-02-07backports: get rid of patch directory structureJohannes Berg
This directory structure was never really managed well, simply get rid of it. With the patch system that only applies patches when the modified file exists, and with a lot of spatches applying everywhere anyway, it's no longer really useful either. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: remove media include patchJohannes Berg
This no longer applies, and it'd probably be a better idea to find why it's needed (and include idr.h in a backport header) or even to send it upstream instead. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: remove some media patchesJohannes Berg
These things should instead be done with an spatch combined with a simple inline that depends on the kernel version. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: remove outdated patch READMEJohannes Berg
Since we're adding mostly spatches, this isn't right - remove it. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: add a simple spatch for skb->wifi_acked[_valid]Johannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: copy linux/bitfield.hJohannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: add USB_SPEED_SUPER_PLUSJohannes Berg
This will cause a compiler warning in some places, but there's no really nice way to work around that. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: add netif_trans_update()Johannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: add ETH_MAX_MTUJohannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: remove 802.15.4 entirelyJohannes Berg
With 802.15.4 code removed, CPTCFG_IEEE802154_6LOWPAN can be inherited from the base kernel and thus cause problems - remove it all. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports-update-manager: add new kernelsJohannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports-update-manager: remove rebuild stepJohannes Berg
This really breaks more than it fixes, afterwards it's no longer possible to compile against those headers (modpost doesn't exist due to the PIC issue.) Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports-update-manager: install kernels one by oneJohannes Berg
That way, the user at least has some feedback that something is happening. Also add the missing try/finally so that the temporary directory is cleaned up. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports-update-manager: remove dead codeJohannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports-update-manager: remove a stray space from a messageJohannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: add compiler-gcc{5,6,7}.hJohannes Berg
They all just include compiler-gccN.h, with the contents taken from the upstream commit merging them all back into compiler-gcc.h. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: include rhashtable only when neededJohannes Berg
This avoids including it when it doesn't exist, on older kernels. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: always disable PIEJohannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: guard genetlink uapi include_nextJohannes Berg
That file didn't always exist, include the non-uapi one where needed. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: remove media, bluetooth, ethernet and 6lowpanJohannes Berg
None of these are maintaining backports anymore, so remove them for now. If anyone's interested, fixes and maintenance is very much welcome. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: adjust bcma of_irq patchJohannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: add sg_pcopy_from_bufferJohannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: add definition for dev_of_node()Arend Van Spriel
Since 4.10-rc1 there is use of dev_of_node() added to cfg80211, which was introduced in 4.1 kernel. So backport the function for older kernels. Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: add CLASS_ATTR_RW/CLASS_ATTR_ROJohannes Berg
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: support class_groupsJohannes Berg
Add a new semantic patch just like for dev_groups to support class_groups. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-07backports: restrict group_attr spatch properlyJohannes Berg
In cases where there are multiple instances of the ATTRIBUTE_GROUPS() macro, or an instance thereof with a .dev_groups that doesn't actually use it (like in devcoredump.c), the spatch wasn't correctly restricted to make sure the two belong together, and thus made some completely wrong modifications, linking an ATTRIBUTE_GROUPS() to a .dev_attrs it didn't belong to at all. Fix this by making sure that the ATTRIBUTE_GROUPS() instance and use for .dev_groups match. This needs python indirection to create the symbol name that ATTRIBUTE_GROUPS() creates in the C code. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: devcoredump: make patch applyJohannes Berg
Just make the patch apply again - the next commits will auto-generate the necessary backport code. Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: disable fils for kernels that don't support skcipherLuca Coelho
It's very hard to backport the skcipher implementation from newer kernels, so disable FILS support for kernels that don't support skcipher natively (< 4.3). Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: empty define for __ro_after_initArend van Spriel
nl80211 now uses section qualifier __ro_after_init. However, this is not available in kernels before v4.6. Neither is the section itself hence adding an empty define in backports. Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com> Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: add backport for nla_memdup()Arend van Spriel
New helper function was introduced in netlink so a backport is needed. Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com> Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: add backport for genl_family_attrbuf() functionArend van Spriel
The function genl_family_attrbuf() was added to the kernel. This backport caries the warning for parallel ops, but it does not assert the genl_mutex as it does not have access to it. Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com> Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06patches: network: add spatch for mtu range checkingArend van Spriel
The mtu range checking was moved into network subsystem resulting in the addition of min_mtu and max_mtu fields in struct net_device. For drivers with a .ndo_change_mtu() which took care of range checking this resulted in removal of the callback. This spatch restores the callback using the values set in {min,max}_mtu fields for kernels before v4.10. Signed-off-by: Julia Lawall <julia.lawall@lip6.fr> Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com> Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: iwlwifi: access dentry name path directly for < 3.12Luca Coelho
In 3.12, new string format types have been added (%pd and %pD), which allows printing a dentry name or file instead of accessing them directly. This is really difficult (if at all feasible) to backport, so we need to keep the old code in the iwlwifi driver for kernels that don't have this new feature. Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: only backport kconfig.h macros for < v3.4Luca Coelho
In kernel v4.9-rc3, config_enabled() was removed, so we can't use it anymore. The IS_BUILTIN macro is undefined then redefined using config_enabled(), which doesn't work anymore. Additionally, it's not necessary to include all these macros if the kernel is >= v3.4. To fix the config_enabled() problem and to avoid backporting unnecessarily, simply gate the backports on < v3.4. Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06git-tracker: add an option to add a Change-Id to commitsJohannes Berg
When the resulting output trees are used with gerrit, it can be useful for patches to have a Change-Id. Add an option to let the script generate them. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backport: remove current_user_ns backportLuca Coelho
In v4.6, the current_user_ns macro was changed to a function, so the ifdef doesn't work. To make things easy, this backport is not needed anymore, since current_user_ns was introduced before v3.0 and we don't support anything earlier than that. Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backport: add nla_put_s64 backport for < v4.7Luca Coelho
The nla_put_s64() function got a new attribute (padding) in v4.7. Add a backport for it. Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06gentree: create Kconfig.local, use it to restrict compatJohannes Berg
Not all the compat code is always necessary, for example code enabled by BPAUTO_CRYPTO_SKCIPHER and by BPAUTO_RHASHTABLE is gated on other symbols being selected. Checking against the Kconfig symbols as it's done right now is wrong though, since the base kernel's Kconfig symbols would be used, selecting, for example, BPAUTO_CRYPTO_SKCIPHER when the base kernel has CONFIG_BT set, but doing that when BT isn't even part of the backport, or when it's disabled in the backport. To fix this, keep track of all the local Kconfig symbols, similar to the .local-symbols files but as a Kconfig.local, and add for each local symbol a symbol that just mirrors the state of the symbol as selected by the user. This effectively allows checking the source of the symbol. Make the relevant backported items use the new BACKPORTED_* symbols for dependencies, instead of the original ones, thus avoiding the above-mentioned "leakage" of kernel symbols. When the symbol isn't even present in the backport, it'll be simply regarded as false by the kconfig system. Reported-by: Jouni Malinen <j@w1.fi> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: add reciprocal_scale() to kernels < 3.14Luca Coelho
This is now needed by the fair queue code. Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06[NOUPSTREAM] backports: add dummy skb_get_hash_perturb() backportLuca Coelho
In kernels < 4.2, we don't have skb_get_hash_perturb(), which is needed by the new fq code. We don't need it in stack-dev, so just add a dummy function that returns 0. Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: genetlink: move GENL_UNS_ADMIN_PERM to uapiLuca Coelho
Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: use codel and fq headers from the kernel sourcesLuca Coelho
We don't need to have a local copy of codel.h anymore, so it can be removed and copied from the kernel sources. Additionally, it was split into 3 different headers, so copy the other 2 as well. Also copy the fq header files, which are used by newer wireless subsystem versions, from the kernel sources. Additionally, remove pkt_sched.h since it is not necessary anymore. Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backport: improve thermal backportLuca Coelho
There were some API changes in the thermal framework in kernel version 4.3 and also earlier in 3.10. Backport what is needed to support older kernels. The 4.3 change is a bit tricky, because it changes the prototypes of some ops. The solution for that is to add hook functions that will intercept the calls from the thermal framework and convert them to calls that the backported driver provides (namely convert unsigned longs to ints). Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: add some scatterlist backports for <3.9 and <3.11Luca Coelho
There were some changes and additions in kernel 3.9 and again in 3.11. Add the needed backports accordingly. Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: add backport for old devcoredumpm (i.e. < 4.7)Luca Coelho
The API has changed in kernel version 4.7. So now we need to backport to this intermediate version if the kernel is > 3.18, but < 4.7. To do so, add wrappers around the old functions and save the functions passed by the caller so that they can be passed by the wrappers. Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: patches: update devcoredump patchArend van Spriel
Due to changes in include/linux/devcoredump.h the patch needs to be updated. Signed-off-by: Arend van Spriel <arend.vanspriel@broadcom.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06gentree: don't throw exceptions when applying patches failLuca Coelho
In commit 4e84c2d807d5 ("backports: move legacy and SmPL patch application into helper"), the functionality was changed by raising exceptions instead of return an error when applying patches fail. This breaks git-tracker, because it expects errors to be returned in such cases, to exit gracefully and print the error messages. Fixes: 4e84c2d807d5 ("backports: move legacy and SmPL patch application into helper") Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: netlink: add 64-bit aligned helper functionsArend van Spriel
This backports the following commits to support nl80211 changes: 089bf1a6a924 ("libnl: add more helpers to align attributes on 64-bit") 35c5845957c7 ("net: Add helpers for 64-bit aligning netlink attributes.") Signed-off-by: Arend van Spriel <arend@broadcom.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backport: fix the backported led_classdev structLuca Coelho
In leds-disabled.h, if any of the three LED flags is disabled in .config (i.e. NEW_LEDS, LEDS_CLASS or LEDS_TRIGGERS), a dummy struct is added. The problem is that this structure doesn't seem to have been updated in a while and, if LEDS_TRIGGERS is not defined, we would get the following errors: /home/luca/iwlwifi/iwlwifi-stack-dev/compat/backport-4.5.c:23:5: error: conflicting types for ‘backport_led_set_brightness_sync’ int led_set_brightness_sync(struct led_classdev *led_cdev, ^ In file included from /home/luca/iwlwifi/iwlwifi-stack-dev/compat/backport-4.5.c:11:0: /home/luca/iwlwifi/iwlwifi-stack-dev/backport-include/linux/leds.h:55:12: note: previous declaration of ‘backport_led_set_brightness_sync’ was here extern int led_set_brightness_sync(struct led_classdev *led_cdev, ^ /home/luca/iwlwifi/iwlwifi-stack-dev/compat/backport-4.5.c: In function ‘backport_led_set_brightness_sync’: /home/luca/iwlwifi/iwlwifi-stack-dev/compat/backport-4.5.c:29:143: warning: comparison of distinct pointer types lacks a cast /home/luca/iwlwifi/iwlwifi-stack-dev/compat/backport-4.5.c:34:14: error: ‘struct backport_led_classdev’ has no member named ‘brightness_set_sync’ if (led_cdev->brightness_set_sync) ^ /home/luca/iwlwifi/iwlwifi-stack-dev/compat/backport-4.5.c:35:18: error: ‘struct backport_led_classdev’ has no member named ‘brightness_set_sync’ return led_cdev->brightness_set_sync(led_cdev, ^ /home/luca/iwlwifi/iwlwifi-stack-dev/compat/backport-4.5.c: At top level: /home/luca/iwlwifi/iwlwifi-stack-dev/compat/backport-4.5.c:39:49: error: conflicting types for ‘backport_led_set_brightness_sync’ In file included from /home/luca/iwlwifi/iwlwifi-stack-dev/compat/backport-4.5.c:11:0: /home/luca/iwlwifi/iwlwifi-stack-dev/backport-include/linux/leds.h:55:12: note: previous declaration of ‘backport_led_set_brightness_sync’ was here extern int led_set_brightness_sync(struct led_classdev *led_cdev, ^ scripts/Makefile.build:258: recipe for target '/home/luca/iwlwifi/iwlwifi-stack-dev/compat/backport-4.5.o' failed To fix this, update the structure accordingly. Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2017-02-06backports: support ullong module_paramEliad Peller
backport this commit to support ullong module params: commit b4210b810e5040f10a30ba56de6c3faab5c49345 Author: Hannes Reinecke <hare@suse.de> Date: Wed Jun 25 15:27:37 2014 +0200 Add module param type 'ullong' Some driver might want to pass in an 64-bit value, so introduce a module param type 'ullong'. Signed-off-by: Hannes Reinecke <hare@suse.de> Signed-off-by: Eliad Peller <eliadx.peller@intel.com> Signed-off-by: Johannes Berg <johannes.berg@intel.com>
generated by cgit v1.2.3 (git 2.0.4) at 2024-04-25 12:31:36 +0000