diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2012-05-29 18:32:37 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2012-05-29 18:32:37 -0700 |
commit | 87a5af24e54857e7b15c1f1b0468512ee65c94e3 (patch) | |
tree | ee346852a0fc27f172a5eb57b6e3c7bf111f2fad /drivers/edac/sb_edac.c | |
parent | 7e5b2db77b05746613516599c916a8cc2e321077 (diff) | |
parent | 0bf09e829dd4b07227ed5a8bc4ac85752a044458 (diff) |
Merge git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-edac
Pull EDAC internal API changes from Mauro Carvalho Chehab:
"This changeset is the first part of a series of patches that fixes the
EDAC sybsystem. On this set, it changes the Kernel EDAC API in order
to properly represent the Intel i3/i5/i7, Xeon 3xxx/5xxx/7xxx, and
Intel E5-xxxx memory controllers.
The EDAC core used to assume that:
- the DRAM chip select pin is directly accessed by the memory
controller
- when multiple channels are used, they're all filled with the
same type of memory.
None of the above premises is true on Intel memory controllers since
2002, when RAMBUS and FB-DIMMs were introduced, and Advanced Memory
Buffer or by some similar technologies hides the direct access to the
DRAM pins.
So, the existing drivers for those chipsets had to lie to the EDAC
core, in general telling that just one channel is filled. That
produces some hard to understand error messages like:
EDAC MC0: CE row 3, channel 0, label "DIMM1": 1 Unknown error(s): memory read error on FATAL area : cpu=0 Err=0008:00c2 (ch=2), addr = 0xad1f73480 => socket=0, Channel=0(mask=2), rank=1
The location information there (row3 channel 0) is completely bogus:
it has no physical meaning, and are just some random values that the
driver uses to talk with the EDAC core. The error actually happened
at CPU socket 0, channel 0, slot 1, but this is not reported anywhere,
as the EDAC core doesn't know anything about the memory layout. So,
only advanced users that know how the EDAC driver works and that tests
their systems to see how DIMMs are mapped can actually benefit for
such error logs.
This patch series fixes the error report logic, in order to allow the
EDAC to expose the memory architecture used by them to the EDAC core.
So, as the EDAC core now understands how the memory is organized, it
can provide an useful report:
EDAC MC0: CE memory read error on DIMM1 (channel:0 slot:1 page:0x364b1b offset:0x600 grain:32 syndrome:0x0 - count:1 area:DRAM err_code:0001:0090 socket:0 channel_mask:1 rank:4)
The location of the DIMM where the error happened is reported by "MC0"
(cpu socket #0), at "channel:0 slot:1" location, and matches the
physical location of the DIMM.
There are two remaining issues not covered by this patch series:
- The EDAC sysfs API will still report bogus values. So,
userspace tools like edac-utils will still use the bogus data;
- Add a new tracepoint-based way to get the binary information
about the errors.
Those are on a second series of patches (also at -next), but will
probably miss the train for 3.5, due to the slow review process."
Fix up trivial conflict (due to spelling correction of removed code) in
drivers/edac/edac_device.c
* git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-edac: (42 commits)
i7core: fix ranks information at the per-channel struct
i5000: Fix the fatal error handling
i5100_edac: Fix a warning when compiled with 32 bits
i82975x_edac: Test nr_pages earlier to save a few CPU cycles
e752x_edac: provide more info about how DIMMS/ranks are mapped
i5000_edac: Fix the logic that retrieves memory information
i5400_edac: improve debug messages to better represent the filled memory
edac: Cleanup the logs for i7core and sb edac drivers
edac: Initialize the dimm label with the known information
edac: Remove the legacy EDAC ABI
x38_edac: convert driver to use the new edac ABI
tile_edac: convert driver to use the new edac ABI
sb_edac: convert driver to use the new edac ABI
r82600_edac: convert driver to use the new edac ABI
ppc4xx_edac: convert driver to use the new edac ABI
pasemi_edac: convert driver to use the new edac ABI
mv64x60_edac: convert driver to use the new edac ABI
mpc85xx_edac: convert driver to use the new edac ABI
i82975x_edac: convert driver to use the new edac ABI
i82875p_edac: convert driver to use the new edac ABI
...
Diffstat (limited to 'drivers/edac/sb_edac.c')
-rw-r--r-- | drivers/edac/sb_edac.c | 212 |
1 files changed, 78 insertions, 134 deletions
diff --git a/drivers/edac/sb_edac.c b/drivers/edac/sb_edac.c index 123204f8e23b..4adaf4b7da99 100644 --- a/drivers/edac/sb_edac.c +++ b/drivers/edac/sb_edac.c @@ -314,8 +314,6 @@ struct sbridge_pvt { struct sbridge_info info; struct sbridge_channel channel[NUM_CHANNELS]; - int csrow_map[NUM_CHANNELS][MAX_DIMMS]; - /* Memory type detection */ bool is_mirrored, is_lockstep, is_close_pg; @@ -487,29 +485,14 @@ static struct pci_dev *get_pdev_slot_func(u8 bus, unsigned slot, } /** - * sbridge_get_active_channels() - gets the number of channels and csrows + * check_if_ecc_is_active() - Checks if ECC is active * bus: Device bus - * @channels: Number of channels that will be returned - * @csrows: Number of csrows found - * - * Since EDAC core needs to know in advance the number of available channels - * and csrows, in order to allocate memory for csrows/channels, it is needed - * to run two similar steps. At the first step, implemented on this function, - * it checks the number of csrows/channels present at one socket, identified - * by the associated PCI bus. - * this is used in order to properly allocate the size of mci components. - * Note: one csrow is one dimm. */ -static int sbridge_get_active_channels(const u8 bus, unsigned *channels, - unsigned *csrows) +static int check_if_ecc_is_active(const u8 bus) { struct pci_dev *pdev = NULL; - int i, j; u32 mcmtr; - *channels = 0; - *csrows = 0; - pdev = get_pdev_slot_func(bus, 15, 0); if (!pdev) { sbridge_printk(KERN_ERR, "Couldn't find PCI device " @@ -523,41 +506,14 @@ static int sbridge_get_active_channels(const u8 bus, unsigned *channels, sbridge_printk(KERN_ERR, "ECC is disabled. Aborting\n"); return -ENODEV; } - - for (i = 0; i < NUM_CHANNELS; i++) { - u32 mtr; - - /* Device 15 functions 2 - 5 */ - pdev = get_pdev_slot_func(bus, 15, 2 + i); - if (!pdev) { - sbridge_printk(KERN_ERR, "Couldn't find PCI device " - "%2x.%02d.%d!!!\n", - bus, 15, 2 + i); - return -ENODEV; - } - (*channels)++; - - for (j = 0; j < ARRAY_SIZE(mtr_regs); j++) { - pci_read_config_dword(pdev, mtr_regs[j], &mtr); - debugf1("Bus#%02x channel #%d MTR%d = %x\n", bus, i, j, mtr); - if (IS_DIMM_PRESENT(mtr)) - (*csrows)++; - } - } - - debugf0("Number of active channels: %d, number of active dimms: %d\n", - *channels, *csrows); - return 0; } -static int get_dimm_config(const struct mem_ctl_info *mci) +static int get_dimm_config(struct mem_ctl_info *mci) { struct sbridge_pvt *pvt = mci->pvt_info; - struct csrow_info *csr; + struct dimm_info *dimm; int i, j, banks, ranks, rows, cols, size, npages; - int csrow = 0; - unsigned long last_page = 0; u32 reg; enum edac_type mode; enum mem_type mtype; @@ -616,6 +572,8 @@ static int get_dimm_config(const struct mem_ctl_info *mci) u32 mtr; for (j = 0; j < ARRAY_SIZE(mtr_regs); j++) { + dimm = EDAC_DIMM_PTR(mci->layers, mci->dimms, mci->n_layers, + i, j, 0); pci_read_config_dword(pvt->pci_tad[i], mtr_regs[j], &mtr); debugf4("Channel #%d MTR%d = %x\n", i, j, mtr); @@ -634,29 +592,15 @@ static int get_dimm_config(const struct mem_ctl_info *mci) pvt->sbridge_dev->mc, i, j, size, npages, banks, ranks, rows, cols); - csr = &mci->csrows[csrow]; - - csr->first_page = last_page; - csr->last_page = last_page + npages - 1; - csr->page_mask = 0UL; /* Unused */ - csr->nr_pages = npages; - csr->grain = 32; - csr->csrow_idx = csrow; - csr->dtype = (banks == 8) ? DEV_X8 : DEV_X4; - csr->ce_count = 0; - csr->ue_count = 0; - csr->mtype = mtype; - csr->edac_mode = mode; - csr->nr_channels = 1; - csr->channels[0].chan_idx = i; - csr->channels[0].ce_count = 0; - pvt->csrow_map[i][j] = csrow; - snprintf(csr->channels[0].label, - sizeof(csr->channels[0].label), + + dimm->nr_pages = npages; + dimm->grain = 32; + dimm->dtype = (banks == 8) ? DEV_X8 : DEV_X4; + dimm->mtype = mtype; + dimm->edac_mode = mode; + snprintf(dimm->label, sizeof(dimm->label), "CPU_SrcID#%u_Channel#%u_DIMM#%u", pvt->sbridge_dev->source_id, i, j); - last_page += npages; - csrow++; } } } @@ -844,11 +788,10 @@ static int get_memory_error_data(struct mem_ctl_info *mci, u8 *socket, long *channel_mask, u8 *rank, - char *area_type) + char **area_type, char *msg) { struct mem_ctl_info *new_mci; struct sbridge_pvt *pvt = mci->pvt_info; - char msg[256]; int n_rir, n_sads, n_tads, sad_way, sck_xch; int sad_interl, idx, base_ch; int interleave_mode; @@ -870,12 +813,10 @@ static int get_memory_error_data(struct mem_ctl_info *mci, */ if ((addr > (u64) pvt->tolm) && (addr < (1LL << 32))) { sprintf(msg, "Error at TOLM area, on addr 0x%08Lx", addr); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } if (addr >= (u64)pvt->tohm) { sprintf(msg, "Error at MMIOH area, on addr 0x%016Lx", addr); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } @@ -892,7 +833,6 @@ static int get_memory_error_data(struct mem_ctl_info *mci, limit = SAD_LIMIT(reg); if (limit <= prv) { sprintf(msg, "Can't discover the memory socket"); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } if (addr <= limit) @@ -901,10 +841,9 @@ static int get_memory_error_data(struct mem_ctl_info *mci, } if (n_sads == MAX_SAD) { sprintf(msg, "Can't discover the memory socket"); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } - area_type = get_dram_attr(reg); + *area_type = get_dram_attr(reg); interleave_mode = INTERLEAVE_MODE(reg); pci_read_config_dword(pvt->pci_sad0, interleave_list[n_sads], @@ -942,7 +881,6 @@ static int get_memory_error_data(struct mem_ctl_info *mci, break; default: sprintf(msg, "Can't discover socket interleave"); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } *socket = sad_interleave[idx]; @@ -957,7 +895,6 @@ static int get_memory_error_data(struct mem_ctl_info *mci, if (!new_mci) { sprintf(msg, "Struct for socket #%u wasn't initialized", *socket); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } mci = new_mci; @@ -973,7 +910,6 @@ static int get_memory_error_data(struct mem_ctl_info *mci, limit = TAD_LIMIT(reg); if (limit <= prv) { sprintf(msg, "Can't discover the memory channel"); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } if (addr <= limit) @@ -1013,7 +949,6 @@ static int get_memory_error_data(struct mem_ctl_info *mci, break; default: sprintf(msg, "Can't discover the TAD target"); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } *channel_mask = 1 << base_ch; @@ -1027,7 +962,6 @@ static int get_memory_error_data(struct mem_ctl_info *mci, break; default: sprintf(msg, "Invalid mirror set. Can't decode addr"); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } } else @@ -1055,7 +989,6 @@ static int get_memory_error_data(struct mem_ctl_info *mci, if (offset > addr) { sprintf(msg, "Can't calculate ch addr: TAD offset 0x%08Lx is too high for addr 0x%08Lx!", offset, addr); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } addr -= offset; @@ -1095,7 +1028,6 @@ static int get_memory_error_data(struct mem_ctl_info *mci, if (n_rir == MAX_RIR_RANGES) { sprintf(msg, "Can't discover the memory rank for ch addr 0x%08Lx", ch_addr); - edac_mc_handle_ce_no_info(mci, msg); return -EINVAL; } rir_way = RIR_WAY(reg); @@ -1409,7 +1341,8 @@ static void sbridge_mce_output_error(struct mem_ctl_info *mci, { struct mem_ctl_info *new_mci; struct sbridge_pvt *pvt = mci->pvt_info; - char *type, *optype, *msg, *recoverable_msg; + enum hw_event_mc_err_type tp_event; + char *type, *optype, msg[256]; bool ripv = GET_BITFIELD(m->mcgstatus, 0, 0); bool overflow = GET_BITFIELD(m->status, 62, 62); bool uncorrected_error = GET_BITFIELD(m->status, 61, 61); @@ -1421,13 +1354,21 @@ static void sbridge_mce_output_error(struct mem_ctl_info *mci, u32 optypenum = GET_BITFIELD(m->status, 4, 6); long channel_mask, first_channel; u8 rank, socket; - int csrow, rc, dimm; - char *area_type = "Unknown"; - - if (ripv) - type = "NON_FATAL"; - else - type = "FATAL"; + int rc, dimm; + char *area_type = NULL; + + if (uncorrected_error) { + if (ripv) { + type = "FATAL"; + tp_event = HW_EVENT_ERR_FATAL; + } else { + type = "NON_FATAL"; + tp_event = HW_EVENT_ERR_UNCORRECTED; + } + } else { + type = "CORRECTED"; + tp_event = HW_EVENT_ERR_CORRECTED; + } /* * According with Table 15-9 of the Intel Architecture spec vol 3A, @@ -1445,19 +1386,19 @@ static void sbridge_mce_output_error(struct mem_ctl_info *mci, } else { switch (optypenum) { case 0: - optype = "generic undef request"; + optype = "generic undef request error"; break; case 1: - optype = "memory read"; + optype = "memory read error"; break; case 2: - optype = "memory write"; + optype = "memory write error"; break; case 3: - optype = "addr/cmd"; + optype = "addr/cmd error"; break; case 4: - optype = "memory scrubbing"; + optype = "memory scrubbing error"; break; default: optype = "reserved"; @@ -1466,13 +1407,13 @@ static void sbridge_mce_output_error(struct mem_ctl_info *mci, } rc = get_memory_error_data(mci, m->addr, &socket, - &channel_mask, &rank, area_type); + &channel_mask, &rank, &area_type, msg); if (rc < 0) - return; + goto err_parsing; new_mci = get_mci_for_node_id(socket); if (!new_mci) { - edac_mc_handle_ce_no_info(mci, "Error: socket got corrupted!"); - return; + strcpy(msg, "Error: socket got corrupted!"); + goto err_parsing; } mci = new_mci; pvt = mci->pvt_info; @@ -1486,45 +1427,39 @@ static void sbridge_mce_output_error(struct mem_ctl_info *mci, else dimm = 2; - csrow = pvt->csrow_map[first_channel][dimm]; - - if (uncorrected_error && recoverable) - recoverable_msg = " recoverable"; - else - recoverable_msg = ""; /* - * FIXME: What should we do with "channel" information on mcelog? - * Probably, we can just discard it, as the channel information - * comes from the get_memory_error_data() address decoding + * FIXME: On some memory configurations (mirror, lockstep), the + * Memory Controller can't point the error to a single DIMM. The + * EDAC core should be handling the channel mask, in order to point + * to the group of dimm's where the error may be happening. */ - msg = kasprintf(GFP_ATOMIC, - "%d %s error(s): %s on %s area %s%s: cpu=%d Err=%04x:%04x (ch=%d), " - "addr = 0x%08llx => socket=%d, Channel=%ld(mask=%ld), rank=%d\n", - core_err_cnt, - area_type, - optype, - type, - recoverable_msg, - overflow ? "OVERFLOW" : "", - m->cpu, - mscod, errcode, - channel, /* 1111b means not specified */ - (long long) m->addr, - socket, - first_channel, /* This is the real channel on SB */ - channel_mask, - rank); + snprintf(msg, sizeof(msg), + "count:%d%s%s area:%s err_code:%04x:%04x socket:%d channel_mask:%ld rank:%d", + core_err_cnt, + overflow ? " OVERFLOW" : "", + (uncorrected_error && recoverable) ? " recoverable" : "", + area_type, + mscod, errcode, + socket, + channel_mask, + rank); debugf0("%s", msg); + /* FIXME: need support for channel mask */ + /* Call the helper to output message */ - if (uncorrected_error) - edac_mc_handle_fbd_ue(mci, csrow, 0, 0, msg); - else - edac_mc_handle_fbd_ce(mci, csrow, 0, msg); + edac_mc_handle_error(tp_event, mci, + m->addr >> PAGE_SHIFT, m->addr & ~PAGE_MASK, 0, + channel, dimm, -1, + optype, msg, m); + return; +err_parsing: + edac_mc_handle_error(tp_event, mci, 0, 0, 0, + -1, -1, -1, + msg, "", m); - kfree(msg); } /* @@ -1683,16 +1618,25 @@ static void sbridge_unregister_mci(struct sbridge_dev *sbridge_dev) static int sbridge_register_mci(struct sbridge_dev *sbridge_dev) { struct mem_ctl_info *mci; + struct edac_mc_layer layers[2]; struct sbridge_pvt *pvt; - int rc, channels, csrows; + int rc; /* Check the number of active and not disabled channels */ - rc = sbridge_get_active_channels(sbridge_dev->bus, &channels, &csrows); + rc = check_if_ecc_is_active(sbridge_dev->bus); if (unlikely(rc < 0)) return rc; /* allocate a new MC control structure */ - mci = edac_mc_alloc(sizeof(*pvt), csrows, channels, sbridge_dev->mc); + layers[0].type = EDAC_MC_LAYER_CHANNEL; + layers[0].size = NUM_CHANNELS; + layers[0].is_virt_csrow = false; + layers[1].type = EDAC_MC_LAYER_SLOT; + layers[1].size = MAX_DIMMS; + layers[1].is_virt_csrow = true; + mci = edac_mc_alloc(sbridge_dev->mc, ARRAY_SIZE(layers), layers, + sizeof(*pvt)); + if (unlikely(!mci)) return -ENOMEM; |