| Message ID | 20180510112616.20508-1-christophe.jaillet@wanadoo.fr |
|---|---|
| State | Accepted, archived |
| Delegated to: | David Miller |
| Headers | show |
| Series | mlxsw: core: Fix an error handling path in 'mlxsw_core_bus_device_register()' | expand |
On Thu, May 10, 2018 at 01:26:16PM +0200, Christophe JAILLET wrote: > Resources are not freed in the reverse order of the allocation. > Labels are also mixed-up. > > Fix it and reorder code and labels in the error handling path of > 'mlxsw_core_bus_device_register()' > > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> For net: Fixes: ef3116e5403e ("mlxsw: spectrum: Register KVD resources with devlink") Reviewed-by: Ido Schimmel <idosch@mellanox.com> Next time, please indicate the tree you're targeting as explained here: https://www.kernel.org/doc/Documentation/networking/netdev-FAQ.txt And include a Fixes line as explained here: https://www.kernel.org/doc/Documentation/process/submitting-patches.rst Thanks!
On Thu, May 10, 2018 at 01:26:16PM +0200, Christophe JAILLET wrote: > Resources are not freed in the reverse order of the allocation. > Labels are also mixed-up. > > Fix it and reorder code and labels in the error handling path of > 'mlxsw_core_bus_device_register()' > > Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> > --- > Please review carefully. This patch is proposed because it triggers one of > my coccinelle scripts. I'm not 100% sure if correct. > Looks correct. The err = mlxsw_driver->resources_register(mlxsw_core); pointer is a pointer to mlxsw_sp_resources_register(). That function doesn't clean up after itself on failure. Ideally, you'd want a matching mlxsw_driver->resources_unregister as well pointer instead of hard coding devlink_resources_unregister(). The error handling would be easier to review if the gotos told you what the did. Right now they're written in "come from" style so the tell you what happened on the line before. if (!foo) goto allocating_foo_failed; Hopefully if someone fixes mlxsw_sp_resources_register() they'll choose better label names. regards, dan carpenter
Hi Dan, I will fix the error path. Regarding the goto label this is the convention in the driver. Thanks, Arkadi
On Thu, May 10, 2018 at 04:08:22PM +0300, Arkadi Sharshevsky wrote: > Hi Dan, > > I will fix the error path. Regarding the goto label this is > the convention in the driver. There is no rule against learning from the past. Or there might be... I don't know all the rules at Mellanox. regards, dan carpenter
From: Ido Schimmel <idosch@idosch.org> Date: Thu, 10 May 2018 14:58:21 +0300 > On Thu, May 10, 2018 at 01:26:16PM +0200, Christophe JAILLET wrote: >> Resources are not freed in the reverse order of the allocation. >> Labels are also mixed-up. >> >> Fix it and reorder code and labels in the error handling path of >> 'mlxsw_core_bus_device_register()' >> >> Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> > > For net: > > Fixes: ef3116e5403e ("mlxsw: spectrum: Register KVD resources with devlink") > Reviewed-by: Ido Schimmel <idosch@mellanox.com> Applied and queued up for -stable, thanks.
diff --git a/drivers/net/ethernet/mellanox/mlxsw/core.c b/drivers/net/ethernet/mellanox/mlxsw/core.c index 93ea56620a24..e13ac3b8dff7 100644 --- a/drivers/net/ethernet/mellanox/mlxsw/core.c +++ b/drivers/net/ethernet/mellanox/mlxsw/core.c @@ -1100,11 +1100,11 @@ int mlxsw_core_bus_device_register(const struct mlxsw_bus_info *mlxsw_bus_info, err_alloc_lag_mapping: mlxsw_ports_fini(mlxsw_core); err_ports_init: - mlxsw_bus->fini(bus_priv); -err_bus_init: if (!reload) devlink_resources_unregister(devlink, NULL); err_register_resources: + mlxsw_bus->fini(bus_priv); +err_bus_init: if (!reload) devlink_free(devlink); err_devlink_alloc:
Resources are not freed in the reverse order of the allocation. Labels are also mixed-up. Fix it and reorder code and labels in the error handling path of 'mlxsw_core_bus_device_register()' Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr> --- Please review carefully. This patch is proposed because it triggers one of my coccinelle scripts. I'm not 100% sure if correct. The script tries to spot wrongly ordered error handling path. It is: @@ identifier l1, l2; @@ if (...) { ... * goto l1; } ... if (...) { ... * goto l2; } ... *l1: ... *l2: ... --- drivers/net/ethernet/mellanox/mlxsw/core.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)