edac: rewrite the sysfs code to use struct device
The EDAC subsystem uses the old struct sysdev approach,
creating all nodes using the raw sysfs API. This is bad,
as the API is deprecated.
As we'll be changing the EDAC API, let's first port the existing
code to struct device.
There's one drawback on this patch: driver-specific sysfs
nodes, used by mpc85xx_edac, amd64_edac and i7core_edac
won't be created anymore. While it would be possible to
also port the device-specific code, that would mix kobj with
struct device, with is not recommended. Also, it is easier and nicer
to move the code to the drivers, instead, as the core can get rid
of some complex logic that just emulates what the device_add()
and device_create_file() already does.
The next patches will convert the driver-specific code to use
the device-specific calls. Then, the remaining bits of the old
sysfs API will be removed.
NOTE: a per-MC bus is required, otherwise devices with more than
one memory controller will hit a bug like the one below:
[ 819.094946] EDAC DEBUG: find_mci_by_dev: find_mci_by_dev()
[ 819.094948] EDAC DEBUG: edac_create_sysfs_mci_device: edac_create_sysfs_mci_device() idx=1
[ 819.094952] EDAC DEBUG: edac_create_sysfs_mci_device: edac_create_sysfs_mci_device(): creating device mc1
[ 819.094967] EDAC DEBUG: edac_create_sysfs_mci_device: edac_create_sysfs_mci_device creating dimm0, located at channel 0 slot 0
[ 819.094984] ------------[ cut here ]------------
[ 819.100142] WARNING: at fs/sysfs/dir.c:481 sysfs_add_one+0xc1/0xf0()
[ 819.107282] Hardware name: S2600CP
[ 819.111078] sysfs: cannot create duplicate filename '/bus/edac/devices/dimm0'
[ 819.119062] Modules linked in: sb_edac(+) edac_core ip6table_filter ip6_tables ebtable_nat ebtables ipt_MASQUERADE iptable_nat nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 xt_state nf_conntrack ipt_REJECT xt_CHECKSUM iptable_mangle iptable_filter ip_tables bridge stp llc sunrpc binfmt_misc dm_mirror dm_region_hash dm_log vhost_net macvtap macvlan tun kvm microcode pcspkr iTCO_wdt iTCO_vendor_support igb i2c_i801 i2c_core sg ioatdma dca sr_mod cdrom sd_mod crc_t10dif ahci libahci isci libsas libata scsi_transport_sas scsi_mod wmi dm_mod [last unloaded: scsi_wait_scan]
[ 819.175748] Pid: 10902, comm: modprobe Not tainted 3.3.0-0.11.el7.v12.2.x86_64 #1
[ 819.184113] Call Trace:
[ 819.186868] [<ffffffff8105adaf>] warn_slowpath_common+0x7f/0xc0
[ 819.193573] [<ffffffff8105aea6>] warn_slowpath_fmt+0x46/0x50
[ 819.200000] [<ffffffff811f53d1>] sysfs_add_one+0xc1/0xf0
[ 819.206025] [<ffffffff811f5cf5>] sysfs_do_create_link+0x135/0x220
[ 819.212944] [<ffffffff811f7023>] ? sysfs_create_group+0x13/0x20
[ 819.219656] [<ffffffff811f5df3>] sysfs_create_link+0x13/0x20
[ 819.226109] [<ffffffff813b04f6>] bus_add_device+0xe6/0x1b0
[ 819.232350] [<ffffffff813ae7cb>] device_add+0x2db/0x460
[ 819.238300] [<ffffffffa0325634>] edac_create_dimm_object+0x84/0xf0 [edac_core]
[ 819.246460] [<ffffffffa0325e18>] edac_create_sysfs_mci_device+0xe8/0x290 [edac_core]
[ 819.255215] [<ffffffffa0322e2a>] edac_mc_add_mc+0x5a/0x2c0 [edac_core]
[ 819.262611] [<ffffffffa03412df>] sbridge_register_mci+0x1bc/0x279 [sb_edac]
[ 819.270493] [<ffffffffa03417a3>] sbridge_probe+0xef/0x175 [sb_edac]
[ 819.277630] [<ffffffff813ba4e8>] ? pm_runtime_enable+0x58/0x90
[ 819.284268] [<ffffffff812f430c>] local_pci_probe+0x5c/0xd0
[ 819.290508] [<ffffffff812f5ba1>] __pci_device_probe+0xf1/0x100
[ 819.297117] [<ffffffff812f5bea>] pci_device_probe+0x3a/0x60
[ 819.303457] [<ffffffff813b1003>] really_probe+0x73/0x270
[ 819.309496] [<ffffffff813b138e>] driver_probe_device+0x4e/0xb0
[ 819.316104] [<ffffffff813b149b>] __driver_attach+0xab/0xb0
[ 819.322337] [<ffffffff813b13f0>] ? driver_probe_device+0xb0/0xb0
[ 819.329151] [<ffffffff813af5d6>] bus_for_each_dev+0x56/0x90
[ 819.335489] [<ffffffff813b0d7e>] driver_attach+0x1e/0x20
[ 819.341534] [<ffffffff813b0980>] bus_add_driver+0x1b0/0x2a0
[ 819.347884] [<ffffffffa0347000>] ? 0xffffffffa0346fff
[ 819.353641] [<ffffffff813b19f6>] driver_register+0x76/0x140
[ 819.359980] [<ffffffff8159f18b>] ? printk+0x51/0x53
[ 819.365524] [<ffffffffa0347000>] ? 0xffffffffa0346fff
[ 819.371291] [<ffffffff812f5896>] __pci_register_driver+0x56/0xd0
[ 819.378096] [<ffffffffa0347054>] sbridge_init+0x54/0x1000 [sb_edac]
[ 819.385231] [<ffffffff8100203f>] do_one_initcall+0x3f/0x170
[ 819.391577] [<ffffffff810bcd2e>] sys_init_module+0xbe/0x230
[ 819.397926] [<ffffffff815bb529>] system_call_fastpath+0x16/0x1b
[ 819.404633] ---[ end trace 1654fdd39556689f ]---
This happens because the bus is not being properly initialized.
Instead of putting the memory sub-devices inside the memory controller,
it is putting everything under the same directory:
$ tree /sys/bus/edac/
/sys/bus/edac/
├── devices
│ ├── all_channel_counts -> ../../../devices/system/edac/mc/mc0/all_channel_counts
│ ├── csrow0 -> ../../../devices/system/edac/mc/mc0/csrow0
│ ├── csrow1 -> ../../../devices/system/edac/mc/mc0/csrow1
│ ├── csrow2 -> ../../../devices/system/edac/mc/mc0/csrow2
│ ├── dimm0 -> ../../../devices/system/edac/mc/mc0/dimm0
│ ├── dimm1 -> ../../../devices/system/edac/mc/mc0/dimm1
│ ├── dimm3 -> ../../../devices/system/edac/mc/mc0/dimm3
│ ├── dimm6 -> ../../../devices/system/edac/mc/mc0/dimm6
│ ├── inject_addrmatch -> ../../../devices/system/edac/mc/mc0/inject_addrmatch
│ ├── mc -> ../../../devices/system/edac/mc
│ └── mc0 -> ../../../devices/system/edac/mc/mc0
├── drivers
├── drivers_autoprobe
├── drivers_probe
└── uevent
On a multi-memory controller system, the names "csrow%d" and "dimm%d"
should be under "mc%d", and not at the main hierarchy level.
So, we need to create a per-MC bus, in order to have its own namespace.
Reviewed-by: Aristeu Rozanski <arozansk@redhat.com>
Cc: Doug Thompson <norsk5@yahoo.com>
Cc: Greg K H <gregkh@linuxfoundation.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
diff --git a/drivers/edac/edac_mc.c b/drivers/edac/edac_mc.c
index 811f09a..61ae346 100644
--- a/drivers/edac/edac_mc.c
+++ b/drivers/edac/edac_mc.c
@@ -218,7 +218,7 @@
unsigned size, tot_dimms = 1, count = 1;
unsigned tot_csrows = 1, tot_channels = 1, tot_errcount = 0;
void *pvt, *p, *ptr = NULL;
- int i, j, err, row, chn, n, len;
+ int i, j, row, chn, n, len;
bool per_rank = false;
BUG_ON(n_layers > EDAC_MAX_LAYERS || n_layers == 0);
@@ -374,15 +374,6 @@
mci->op_state = OP_ALLOC;
INIT_LIST_HEAD(&mci->grp_kobj_list);
- /*
- * Initialize the 'root' kobj for the edac_mc controller
- */
- err = edac_mc_register_sysfs_main_kobj(mci);
- if (err) {
- kfree(mci);
- return NULL;
- }
-
/* at this point, the root kobj is valid, and in order to
* 'free' the object, then the function:
* edac_mc_unregister_sysfs_main_kobj() must be called
@@ -403,7 +394,7 @@
{
debugf1("%s()\n", __func__);
- edac_mc_unregister_sysfs_main_kobj(mci);
+ edac_unregister_sysfs(mci);
/* free the mci instance memory here */
kfree(mci);