[PATCH 10/10] EDAC/mc: Remove per layer counters

From: Robert Richter
Date: Wed Nov 27 2019 - 16:55:20 EST


Looking at how mci->{ue,ce}_per_layer[EDAC_MAX_LAYERS] is used, it
turns out that only the leaves in the memory hierarchy are consumed
(in sysfs), but not the intermediate layers, e.g.:

count = dimm->mci->ce_per_layer[dimm->mci->n_layers-1][dimm->idx];

These unused counters only add complexity, remove them. The error
counter values are directly stored in struct dimm_info now.

Signed-off-by: Robert Richter <rrichter@xxxxxxxxxxx>
---
drivers/edac/edac_mc.c | 65 +++++++++---------------------------
drivers/edac/edac_mc_sysfs.c | 20 +++++------
include/linux/edac.h | 4 ++-
3 files changed, 26 insertions(+), 63 deletions(-)

diff --git a/drivers/edac/edac_mc.c b/drivers/edac/edac_mc.c
index 5ea834fceb50..3d43b9fe8171 100644
--- a/drivers/edac/edac_mc.c
+++ b/drivers/edac/edac_mc.c
@@ -445,11 +445,9 @@ struct mem_ctl_info *edac_mc_alloc(unsigned int mc_num,
{
struct mem_ctl_info *mci;
struct edac_mc_layer *layer;
- u32 *ce_per_layer[EDAC_MAX_LAYERS], *ue_per_layer[EDAC_MAX_LAYERS];
- unsigned int idx, size, tot_dimms = 1, count = 1;
- unsigned int tot_csrows = 1, tot_channels = 1, tot_errcount = 0;
+ unsigned int idx, size, tot_dimms = 1;
+ unsigned int tot_csrows = 1, tot_channels = 1;
void *pvt, *ptr = NULL;
- int i;
bool per_rank = false;

if (WARN_ON(n_layers > EDAC_MAX_LAYERS || n_layers == 0))
@@ -476,19 +474,10 @@ struct mem_ctl_info *edac_mc_alloc(unsigned int mc_num,
* stringent as what the compiler would provide if we could simply
* hardcode everything into a single struct.
*/
- mci = edac_align_ptr(&ptr, sizeof(*mci), 1);
- layer = edac_align_ptr(&ptr, sizeof(*layer), n_layers);
- for (i = 0; i < n_layers; i++) {
- count *= layers[i].size;
- edac_dbg(4, "errcount layer %d size %d\n", i, count);
- ce_per_layer[i] = edac_align_ptr(&ptr, sizeof(u32), count);
- ue_per_layer[i] = edac_align_ptr(&ptr, sizeof(u32), count);
- tot_errcount += 2 * count;
- }
-
- edac_dbg(4, "allocating %d error counters\n", tot_errcount);
- pvt = edac_align_ptr(&ptr, sz_pvt, 1);
- size = ((unsigned long)pvt) + sz_pvt;
+ mci = edac_align_ptr(&ptr, sizeof(*mci), 1);
+ layer = edac_align_ptr(&ptr, sizeof(*layer), n_layers);
+ pvt = edac_align_ptr(&ptr, sz_pvt, 1);
+ size = ((unsigned long)pvt) + sz_pvt;

edac_dbg(1, "allocating %u bytes for mci data (%d %s, %d csrows/channels)\n",
size,
@@ -504,10 +493,6 @@ struct mem_ctl_info *edac_mc_alloc(unsigned int mc_num,
* rather than an imaginary chunk of memory located at address 0.
*/
layer = (struct edac_mc_layer *)(((char *)mci) + ((unsigned long)layer));
- for (i = 0; i < n_layers; i++) {
- mci->ce_per_layer[i] = (u32 *)((char *)mci + ((unsigned long)ce_per_layer[i]));
- mci->ue_per_layer[i] = (u32 *)((char *)mci + ((unsigned long)ue_per_layer[i]));
- }
pvt = sz_pvt ? (((char *)mci) + ((unsigned long)pvt)) : NULL;

/* setup index and various internal pointers */
@@ -949,48 +934,28 @@ static void edac_inc_ce_error(struct edac_raw_error_desc *e)
{
struct mem_ctl_info *mci = error_desc_to_mci(e);
int pos[EDAC_MAX_LAYERS] = { e->top_layer, e->mid_layer, e->low_layer };
- int i, index = 0;
+ struct dimm_info *dimm = edac_get_dimm(mci, pos[0], pos[1], pos[2]);

mci->ce_mc += e->error_count;

- if (pos[0] < 0) {
+ if (dimm)
+ dimm->ce_count += e->error_count;
+ else
mci->ce_noinfo_count += e->error_count;
- return;
- }
-
- for (i = 0; i < mci->n_layers; i++) {
- if (pos[i] < 0)
- break;
- index += pos[i];
- mci->ce_per_layer[i][index] += e->error_count;
-
- if (i < mci->n_layers - 1)
- index *= mci->layers[i + 1].size;
- }
}

static void edac_inc_ue_error(struct edac_raw_error_desc *e)
{
struct mem_ctl_info *mci = error_desc_to_mci(e);
int pos[EDAC_MAX_LAYERS] = { e->top_layer, e->mid_layer, e->low_layer };
- int i, index = 0;
+ struct dimm_info *dimm = edac_get_dimm(mci, pos[0], pos[1], pos[2]);

mci->ue_mc += e->error_count;

- if (pos[0] < 0) {
+ if (dimm)
+ dimm->ue_count += e->error_count;
+ else
mci->ue_noinfo_count += e->error_count;
- return;
- }
-
- for (i = 0; i < mci->n_layers; i++) {
- if (pos[i] < 0)
- break;
- index += pos[i];
- mci->ue_per_layer[i][index] += e->error_count;
-
- if (i < mci->n_layers - 1)
- index *= mci->layers[i + 1].size;
- }
}

static void edac_ce_error(struct edac_raw_error_desc *e)
@@ -1137,7 +1102,7 @@ void edac_mc_handle_error(const enum hw_event_mc_err_type type,
/*
* Check if the event report is consistent and if the memory
* location is known. If it is known, the DIMM(s) label info
- * will be filled and the per-layer error counters will be
+ * will be filled and the DIMM's error counters will be
* incremented.
*/
for (i = 0; i < mci->n_layers; i++) {
diff --git a/drivers/edac/edac_mc_sysfs.c b/drivers/edac/edac_mc_sysfs.c
index 0367554e7437..8682df2f7f4f 100644
--- a/drivers/edac/edac_mc_sysfs.c
+++ b/drivers/edac/edac_mc_sysfs.c
@@ -556,10 +556,8 @@ static ssize_t dimmdev_ce_count_show(struct device *dev,
char *data)
{
struct dimm_info *dimm = to_dimm(dev);
- u32 count;

- count = dimm->mci->ce_per_layer[dimm->mci->n_layers-1][dimm->idx];
- return sprintf(data, "%u\n", count);
+ return sprintf(data, "%u\n", dimm->ce_count);
}

static ssize_t dimmdev_ue_count_show(struct device *dev,
@@ -567,10 +565,8 @@ static ssize_t dimmdev_ue_count_show(struct device *dev,
char *data)
{
struct dimm_info *dimm = to_dimm(dev);
- u32 count;

- count = dimm->mci->ue_per_layer[dimm->mci->n_layers-1][dimm->idx];
- return sprintf(data, "%u\n", count);
+ return sprintf(data, "%u\n", dimm->ue_count);
}

/* dimm/rank attribute files */
@@ -666,7 +662,9 @@ static ssize_t mci_reset_counters_store(struct device *dev,
const char *data, size_t count)
{
struct mem_ctl_info *mci = to_mci(dev);
- int cnt, row, chan, i;
+ struct dimm_info *dimm;
+ int row, chan;
+
mci->ue_mc = 0;
mci->ce_mc = 0;
mci->ue_noinfo_count = 0;
@@ -682,11 +680,9 @@ static ssize_t mci_reset_counters_store(struct device *dev,
ri->channels[chan]->ce_count = 0;
}

- cnt = 1;
- for (i = 0; i < mci->n_layers; i++) {
- cnt *= mci->layers[i].size;
- memset(mci->ce_per_layer[i], 0, cnt * sizeof(u32));
- memset(mci->ue_per_layer[i], 0, cnt * sizeof(u32));
+ mci_for_each_dimm(mci, dimm) {
+ dimm->ue_count = 0;
+ dimm->ce_count = 0;
}

mci->start_time = jiffies;
diff --git a/include/linux/edac.h b/include/linux/edac.h
index d441026eeb1c..76e2c099463b 100644
--- a/include/linux/edac.h
+++ b/include/linux/edac.h
@@ -383,6 +383,9 @@ struct dimm_info {
unsigned int csrow, cschannel; /* Points to the old API data */

u16 smbios_handle; /* Handle for SMBIOS type 17 */
+
+ u32 ce_count;
+ u32 ue_count;
};

/**
@@ -558,7 +561,6 @@ struct mem_ctl_info {
*/
u32 ce_noinfo_count, ue_noinfo_count;
u32 ue_mc, ce_mc;
- u32 *ce_per_layer[EDAC_MAX_LAYERS], *ue_per_layer[EDAC_MAX_LAYERS];

struct completion complete;

--
2.20.1