- 21 6月, 2017 13 次提交
-
-
由 Icenowy Zheng 提交于
The EPHY control part of the EMAC syscon register has a bit called CLK_SEL. On the datasheet it says that if it's 0 the EPHY clock is 25MHz and if it's 1 the clock is 24MHz. However, according to the datasheets, no Allwinner SoC with EPHY has any extra xtal input pins for the EPHY, and the system xtal is 24MHz. That means the EPHY is not possible to get a 25MHz xtal input, and thus the frequency can only be 24MHz. It doesn't matter on H3 as the default value of H3 is 24MHz, however on V3s the default value is wrongly set to 25MHz, which prevented the EPHY from working properly. Force the EPHY clock frequency to 24MHz. Signed-off-by: NIcenowy Zheng <icenowy@aosc.io> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Icenowy Zheng 提交于
Allwinner V3s SoC has a syscon like the one in H3. Add its compatible string. Signed-off-by: NIcenowy Zheng <icenowy@aosc.io> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Icenowy Zheng 提交于
Allwinner V3s SoC has a Ethernet MAC like the one in Allwinner H3, but have no external MII capability. That means that it can only use the EPHY and cannot do Gbps transmission. Add binding for it. Signed-off-by: NIcenowy Zheng <icenowy@aosc.io> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 David S. Miller 提交于
Lucas Bates says: ==================== net: Introduction of the tc tests Apologies for sending this as one big patch. I've been sitting on this a little too long, but it's ready and I wanted to get it out. There are a limited number of tests to start - I plan to add more on a regular basis. ==================== Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Lucas Bates 提交于
Add the beginnings of a testsuite for tc functionality in the kernel. These are a series of unit tests that use the tc executable and verify the success of those commands by checking both the exit codes and the output from tc's 'show' operation. To run the tests: # cd tools/testing/selftests/tc-testing # sudo ./tdc.py You can specify the tc executable to use with the -p argument on the command line or editing the 'TC' variable in tdc_config.py. Refer to the README for full details on how to run. The initial complement of test cases are limited mostly to tc actions. Test cases are most welcome; see the creating-testcases subdirectory for help in creating them. Signed-off-by: NLucas Bates <lucasb@mojatatu.com> Signed-off-by: NJamal Hadi Salim <jhs@mojatatu.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 David S. Miller 提交于
Yuval Mintz says: ==================== qed*: RDMA and infrastructure for iWARP This series focuses on RDMA in general with emphasis on required changes toward adding iWARP support. The vast majority of the changes introduced are in qed/qede, with a couple of small changes to qedr [mentioned below]. The infrastructure changes: - Patch #1 adds the ability to pass PBL memory externally for a newly created chain. - Patches #4, #5 rename qede_roce.[ch] into qede_rdma.[ch] + change prefixes from _roce_ to _rdma_, as the API between qede and qedr is agnostic to the variant of the RDMA protocol used. These patches also touch qedr [basically to align it with the renaming, nothing more]. - Patch #7 replaces the current SPQ async mechanism into serving registered callbacks [before adding iWARP which would add another client in need of this sort of functionallity]. The non-infrastrucutre changes: - Patches #2, #3 contain DCB-related changes to better align RDMA with configured DCB. - Patch #6 contains a minor [mostly theoretical fix] to release flow. Changes from previous versions ------------------------------ - V4: This is actually a repost of V3 due to some confusion regarding the sent cover-letter - V3: Add commit log message in #4 indicating change in header inclusion - V2: Add several inclusion into qede_rdma.h to have proper declarations of all variable types used in it ==================== Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Michal Kalderon 提交于
Whenever firmware indicates that there's an async indication it needs to handle, there's a switch-case where the right functionality is called based on function's personality and information. Before iWARP is added [as yet another client], switch over the SPQ into a callback-registered mechanism, allowing registration of the relevant event-processing logic based on the function's personality. This allows us to tidy the code by removing protocol-specifics from a common file. Signed-off-by: NMichal Kalderon <Michal.Kalderon@cavium.com> Signed-off-by: NYuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Michal Kalderon 提交于
Driver needs to wait for all resources to return from FW before it can send the FUNC_CLOSE ramrod. Signed-off-by: NMichal Kalderon <Michal.Kalderon@cavium.com> Signed-off-by: NYuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Michal Kalderon 提交于
Rename the functions common to both iWARP and RoCE to have a prefix of _rdma_ instead of _roce_. Signed-off-by: NMichal Kalderon <Michal.Kalderon@cavium.com> Signed-off-by: NYuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Michal Kalderon 提交于
Once we have iWARP support, the qede portion of the qedr<->qede would serve all the RDMA protocols - so rename the file to be appropriate to its function. While we're at it, we're also moving a couple of inclusions to it into .h files and adding includes to make sure it contains all type definitions it requires. Signed-off-by: NMichal Kalderon <Michal.Kalderon@cavium.com> Signed-off-by: NYuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Mintz, Yuval 提交于
If DCBx update occurs while QPs are open, stop sending edpms until all QPs are closed. Signed-off-by: NYuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Mintz, Yuval 提交于
Configure device according to DCBx results so that EDPMs made by RoCE would honor flow-control. Signed-off-by: NYuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Mintz, Yuval 提交于
iWARP would require the chains to allocate/free their PBL memory independently, so add the infrastructure to provide it externally. Signed-off-by: NYuval Mintz <Yuval.Mintz@cavium.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
- 20 6月, 2017 2 次提交
-
-
由 Ivan Delalande 提交于
Replace first padding in the tcp_md5sig structure with a new flag field and address prefix length so it can be specified when configuring a new key for TCP MD5 signature. The tcpm_flags field will only be used if the socket option is TCP_MD5SIG_EXT to avoid breaking existing programs, and tcpm_prefixlen only when the TCP_MD5SIG_FLAG_PREFIX flag is set. Signed-off-by: NBob Gilligan <gilligan@arista.com> Signed-off-by: NEric Mowat <mowat@arista.com> Signed-off-by: NIvan Delalande <colona@arista.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Ivan Delalande 提交于
This allows the keys used for TCP MD5 signature to be used for whole range of addresses, specified with a prefix length, instead of only one address as it currently is. Signed-off-by: NBob Gilligan <gilligan@arista.com> Signed-off-by: NEric Mowat <mowat@arista.com> Signed-off-by: NIvan Delalande <colona@arista.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
- 19 6月, 2017 7 次提交
-
-
由 Raju Rangoju 提交于
During the module initialisation there is a possible race (basically race between uld and lld) where neither the uld nor lld notifies the uP about where to route the ctrl queue completions. LLD skips notifying uP as the rdma queues were not created by then (will leave it to ULD to notify the uP). As the ULD comes up, it also skips notifying the uP as the flag FULL_INIT_DONE is not set yet (ULD assumes that the interface is not up yet). Consequently, this race between uld and lld leaves uP unnotified about where to send the ctrl queue completions to, leading to iwarp RI_RES WR failure. Here is the race: CPU 0 CPU1 - allocates nic rx queus - t4_sge_alloc_ctrl_txq() (if rdma rsp queues exists, tell uP to route ctrl queue compl to rdma rspq) - acquires the mutex_lock - allocates rdma response queues - if FULL_INIT_DONE set, tell uP to route ctrl queue compl to rdma rspq - relinquishes mutex_lock - acquires the mutex_lock - enable_rx() - set FULL_INIT_DONE - relinquishes mutex_lock This patch fixes the above issue. Fixes: e7519f99('cxgb4: avoid enabling napi twice to the same queue') Signed-off-by: NRaju Rangoju <rajur@chelsio.com> Acked-by: NSteve Wise <swise@opengridcomputing.com> CC: Stable <stable@vger.kernel.org> # 4.9+ Signed-off-by: NGanesh Goudar <ganeshgr@chelsio.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Ganesh Goudar 提交于
Add 0x6082, 0x6083 and 0x6084 T6 device id's Signed-off-by: NGanesh Goudar <ganeshgr@chelsio.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Pablo Cascón 提交于
Add general use per-vNIC mailbox area and use it for VLAN filtering support. Initially proto is hardcoded to 802.1q. Signed-off-by: NPablo Cascón <pablo.cascon@netronome.com> Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Ganesh Goudar 提交于
Avoid NULL dereference in setup_sge_queues() when the adapter is in non offload mode. Fixes: 0fbc81b3 ('chcr/cxgb4i/cxgbit/RDMA/cxgb4: Allocate resources dynamically for all cxgb4 ULD's') Signed-off-by: NGanesh Goudar <ganeshgr@chelsio.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Prasad Kanneganti 提交于
Each Octeon output ring can DMA packets to host memory in two modes: info- pointer mode and buffer-pointer-only mode. In info-pointer mode, Octeon takes two buffer pointers for each packet and places the length of the packet along with specified number of bytes from the beginning of the packet into one buffer and the rest of the packet in a separate buffer. In buffer-pointer-only mode, Octeon takes single buffer pointer and places the length of the packet at the beginning of the buffer followed by the packet data. This patch switches all Octeon output rings from info-pointer mode to buffer-pointer-only mode. This results in fewer DMA setups and cache line snoops. Signed-off-by: NPrasad Kanneganti <pkanneganti@cavium.com> Signed-off-by: NDerek Chickles <derek.chickles@cavium.com> Signed-off-by: NSatanand Burla <satananda.burla@cavium.com> Signed-off-by: NFelix Manlunas <felix.manlunas@cavium.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Christos Gkekas 提交于
Variable opt in pptp_release() is set but never used, thus needs to be removed. Signed-off-by: NChristos Gkekas <chris.gekas@gmail.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Prasad Kanneganti 提交于
Add implementation to support ethtool -K ethX rx-vlan-filter on/off. Rename OCTNET_CMD_ENABLE_VLAN_FILTER command to OCTNET_CMD_VLAN_FILTER_CTL and add OCTNET_CMD_VLAN_FILTER_ENABLE and OCTNET_CMD_VLAN_FILTER_DISABLE parameters so that it can be used to enable or disable the filter. Signed-off-by: NPrasad Kanneganti <prasad.kanneganti@cavium.com> Signed-off-by: NDerek Chickles <derek.chickles@cavium.com> Signed-off-by: NFelix Manlunas <felix.manlunas@cavium.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
- 18 6月, 2017 18 次提交
-
-
由 Florian Fainelli 提交于
After commit 6d3c8c0d ("net: dsa: Remove master_netdev and use dst->cpu_dp->netdev") and a29342e7 ("net: dsa: Associate slave network device with CPU port") we would be seeing NULL pointer dereferences when accessing dst->cpu_dp->netdev too early. In the legacy code, we actually know early in advance the master network device, so pass it down to the relevant functions. Fixes: 6d3c8c0d ("net: dsa: Remove master_netdev and use dst->cpu_dp->netdev") Fixes: a29342e7 ("net: dsa: Associate slave network device with CPU port") Reported-by: NJason Cobham <jcobham@questertangent.com> Tested-by: NJason Cobham <jcobham@questertangent.com> Signed-off-by: NFlorian Fainelli <f.fainelli@gmail.com> Reviewed-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Dave Watson 提交于
Missing crypto deps for some platforms. Default to n for new module. config: m68k-amcore_defconfig (attached as .config) compiler: m68k-linux-gcc (GCC) 4.9.0 make.cross ARCH=m68k All errors (new ones prefixed by >>): net/built-in.o: In function `tls_set_sw_offload': >> (.text+0x732f8): undefined reference to `crypto_alloc_aead' net/built-in.o: In function `tls_set_sw_offload': >> (.text+0x7333c): undefined reference to `crypto_aead_setkey' net/built-in.o: In function `tls_set_sw_offload': >> (.text+0x73354): undefined reference to `crypto_aead_setauthsize' Reported-by: Nkbuild test robot <fengguang.wu@intel.com> Signed-off-by: NDave Watson <davejwatson@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 David S. Miller 提交于
Wei Wang says: ==================== remove dst garbage collector logic The current mechanism of dst release is a bit complicated. It is because the users of dst get divided into 2 situations: 1. Most users take the reference count when using a dst and release the reference count when done. 2. Exceptional users like IPv4/IPv6/decnet/xfrm routing code do not take reference count when referencing to a dst due to some histotic reasons. Due to those exceptional use cases in 2, reference count being 0 is not an adequate evidence to indicate that no user is using this dst. So users in 1 can't free the dst simply based on reference count being 0 because users in 2 might still hold reference to it. Instead, a dst garbage list is needed to hold the dst entries that already get removed by the users in 2 but are still held by users in 1. And a periodic garbage collector task is run to check all the dst entries in the list to see if the users in 1 have released the reference to those dst entries. If so, the dst is now ready to be freed. This logic introduces unnecessary complications in the dst code which makes it hard to understand and to debug. In order to get rid of the whole dst garbage collector (gc) and make the dst code more unified and simplified, we can make the users in 2 also take reference count on the dst and release it properly when done. This way, dst can be safely freed once the refcount drops to 0 and no gc thread is needed anymore. This patch series' target is to completely get rid of dst gc logic and free dst based on reference count only. Patch 1-3 are preparation patches to do some cleanup/improvement on the existing code to make later work easier. Patch 4-21 are real implementations. In these patches, a temporary flag DST_NOGC is used to help transition those exceptional users one by one. Once every component is transitioned, this temporary flag is removed. By the end of this patch series, all dst are refcounted when being used and released when done. And dst will be freed when its refcount drops to 0. No dst gc task is running anymore. Note: This patch series depends on the decnet fix that was sent right before: "decnet: always not take dst->__refcnt when inserting dst into hash table" v2: add curly braces in udp_v4/6_early_demux() in patch 02 add EXPORT_SYMBOL() for dst_dev_put() in patch 05 ==================== Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
This patch is meant to add a debug warning on the situation where dst is being held during its destroy phase. This could potentially cause double free issue on the dst. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
As some dst flags are removed, reorder the dst flags to fill in the blanks. Note: these flags are not exposed into user space. So it is safe to reorder. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
DST_NOCACHE flag check has been removed from dst_release() and dst_hold_safe() in a previous patch because all the dst are now ref counted properly and can be released based on refcnt only. Looking at the rest of the DST_NOCACHE use, all of them can now be removed or replaced with other checks. So this patch gets rid of all the DST_NOCACHE usage and remove this flag completely. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
Now that all the components have been changed to release dst based on refcnt only and not depend on dst gc anymore, we can remove the temporary flag DST_NOGC. Note that we also need to remove the DST_NOCACHE check in dst_release() and dst_hold_safe() because now all the dst are released based on refcnt and behaves as DST_NOCACHE. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
This patch removes all dst gc related code and all the dst free functions Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
struct dn_route is inserted into dn_rt_hash_table but no dst->__refcnt is taken. This patch makes sure the dn_rt_hash_table's reference to the dst is ref counted. As the dst is always ref counted properly, we can safely mark DST_NOGC flag so dst_release() will release dst based on refcnt only. And dst gc is no longer needed and all dst_free() or its related function calls should be replaced with dst_release() or dst_release_immediate(). And dst_dev_put() is called when removing dst from the hash table to release the reference on dst->dev before we lose pointer to it. Also, correct the logic in dn_dst_check_expire() and dn_dst_gc() to check dst->__refcnt to be > 1 to indicate it is referenced by other users. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
During the creation of xfrm_dst bundle, always take ref count when allocating the dst. This way, xfrm_bundle_create() will form a linked list of dst with dst->child pointing to a ref counted dst child. And the returned dst pointer is also ref counted. This makes the link from the flow cache to this dst now ref counted properly. As the dst is always ref counted properly, we can safely mark DST_NOGC flag so dst_release() will release dst based on refcnt only. And dst gc is no longer needed and all dst_free() and its related function calls should be replaced with dst_release() or dst_release_immediate(). The special handling logic for dst->child in dst_destroy() can be replaced with a simple dst_release_immediate() call on the child to release the whole list linked by dst->child pointer. Previously used DST_NOHASH flag is not needed anymore as well. The reason that DST_NOHASH is used in the existing code is mainly to prevent the dst inserted in the fib tree to be wrongly destroyed during the deletion of the xfrm_dst bundle. So in the existing code, DST_NOHASH flag is marked in all the dst children except the one which is in the fib tree. However, with this patch series to remove dst gc logic and release dst only based on ref count, it is safe to release all the children from a xfrm_dst bundle as long as the dst children are all ref counted properly which is already the case in the existing code. So, this patch removes the use of DST_NOHASH flag. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
icmp6 dst route is currently ref counted during creation and will be freed by user during its call of dst_release(). So no need of a garbage collector for it. Remove all icmp6 dst garbage collector related code. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
With the previous preparation patches, we are ready to get rid of the dst gc operation in ipv6 code and release dst based on refcnt only. So this patch adds DST_NOGC flag for all IPv6 dst and remove the calls to dst_free() and its related functions. At this point, all dst created in ipv6 code do not use the dst gc anymore and will be destroyed at the point when refcnt drops to 0. Also, as icmp6 dst route is refcounted during creation and will be freed by user during its call of dst_release(), there is no need to add this dst to the icmp6 gc list as well. Instead, we need to add it into uncached list so that when a NETDEV_DOWN/NETDEV_UNREGISRER event comes, we can properly go through these icmp6 dst as well and release the net device properly. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
Similar as ipv4, ipv6 path also needs to call dst_hold_safe() when necessary to avoid double free issue on the dst. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
As the intend of this patch series is to completely remove dst gc, we need to call dst_dev_put() to release the reference to dst->dev when removing routes from fib because we won't keep the gc list anymore and will lose the dst pointer right after removing the routes. Without the gc list, there is no way to find all the dst's that have dst->dev pointing to the going-down dev. Hence, we are doing dst_dev_put() immediately before we lose the last reference of the dst from the routing code. The next dst_check() will trigger a route re-lookup to find another route (if there is any). Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
In IPv6 routing code, struct rt6_info is created for each static route and RTF_CACHE route and inserted into fib6 tree. In both cases, dst ref count is not taken. As explained in the previous patch, this leads to the need of the dst garbage collector. This patch holds ref count of dst before inserting the route into fib6 tree and properly releases the dst when deleting it from the fib6 tree as a preparation in order to fully get rid of dst gc later. Also, correct fib6_age() logic to check dst->__refcnt to be 1 to indicate no user is referencing the dst. And remove dst_hold() in vrf_rt6_create() as ip6_dst_alloc() already puts dst->__refcnt to 1. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
With the previous preparation patches, we are ready to get rid of the dst gc operation in ipv4 code and release dst based on refcnt only. So this patch adds DST_NOGC flag for all IPv4 dst and remove the calls to dst_free(). At this point, all dst created in ipv4 code do not use the dst gc anymore and will be destroyed at the point when refcnt drops to 0. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
This patch checks all the calls to dst_hold()/skb_dst_force()/dst_clone()/dst_use() to see if dst_hold_safe() is needed to avoid double free issue if dst gc is removed and dst_release() directly destroys dst when dst->__refcnt drops to 0. In tx path, TCP hold sk->sk_rx_dst ref count and also hold sock_lock(). UDP and other similar protocols always hold refcount for skb->_skb_refdst. So both paths seem to be safe. In rx path, as it is lockless and skb_dst_set_noref() is likely to be used, dst_hold_safe() should always be used when trying to hold dst. In the routing code, if dst is held during an rcu protected session, it is necessary to call dst_hold_safe() as the current dst might be in its rcu grace period. Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-
由 Wei Wang 提交于
As the intend of this patch series is to completely remove dst gc, we need to call dst_dev_put() to release the reference to dst->dev when removing routes from fib because we won't keep the gc list anymore and will lose the dst pointer right after removing the routes. Without the gc list, there is no way to find all the dst's that have dst->dev pointing to the going-down dev. Hence, we are doing dst_dev_put() immediately before we lose the last reference of the dst from the routing code. The next dst_check() will trigger a route re-lookup to find another route (if there is any). Signed-off-by: NWei Wang <weiwan@google.com> Acked-by: NMartin KaFai Lau <kafai@fb.com> Signed-off-by: NDavid S. Miller <davem@davemloft.net>
-