6.4-stable review patch. If anyone has any objections, please let me know.
------------------
From: Qiuxu Zhuo qiuxu.zhuo@intel.com
[ Upstream commit c545f5e412250555bd4e717d062b117f20bab418 ]
Some Sapphire Rapids workstations' absent memory controllers still appear as PCIe devices that fool the i10nm_edac driver and result in "shift exponent -66 is negative" call traces from skx_get_dimm_info().
Skip the absent memory controllers to avoid the call traces.
Reported-by: Kai-Heng Feng kai.heng.feng@canonical.com Closes: https://lore.kernel.org/linux-edac/CAAd53p41Ku1m1rapeqb1xtD+kKuk+BaUW=dumuoF... Tested-by: Kai-Heng Feng kai.heng.feng@canonical.com Reported-by: Koba Ko koba.ko@canonical.com Closes: https://lore.kernel.org/linux-edac/SA1PR11MB71305B71CCCC3D9305835202892AA@SA... Tested-by: Koba Ko koba.ko@canonical.com Fixes: d4dc89d069aa ("EDAC, i10nm: Add a driver for Intel 10nm server processors") Signed-off-by: Qiuxu Zhuo qiuxu.zhuo@intel.com Signed-off-by: Tony Luck tony.luck@intel.com Link: https://lore.kernel.org/r/20230710013232.59712-1-qiuxu.zhuo@intel.com Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/edac/i10nm_base.c | 54 +++++++++++++++++++++++++++++++++++---- 1 file changed, 49 insertions(+), 5 deletions(-)
diff --git a/drivers/edac/i10nm_base.c b/drivers/edac/i10nm_base.c index a897b6aff3686..349ff6cfb3796 100644 --- a/drivers/edac/i10nm_base.c +++ b/drivers/edac/i10nm_base.c @@ -658,13 +658,49 @@ static struct pci_dev *get_ddr_munit(struct skx_dev *d, int i, u32 *offset, unsi return mdev; }
+/** + * i10nm_imc_absent() - Check whether the memory controller @imc is absent + * + * @imc : The pointer to the structure of memory controller EDAC device. + * + * RETURNS : true if the memory controller EDAC device is absent, false otherwise. + */ +static bool i10nm_imc_absent(struct skx_imc *imc) +{ + u32 mcmtr; + int i; + + switch (res_cfg->type) { + case SPR: + for (i = 0; i < res_cfg->ddr_chan_num; i++) { + mcmtr = I10NM_GET_MCMTR(imc, i); + edac_dbg(1, "ch%d mcmtr reg %x\n", i, mcmtr); + if (mcmtr != ~0) + return false; + } + + /* + * Some workstations' absent memory controllers still + * appear as PCIe devices, misleading the EDAC driver. + * By observing that the MMIO registers of these absent + * memory controllers consistently hold the value of ~0. + * + * We identify a memory controller as absent by checking + * if its MMIO register "mcmtr" == ~0 in all its channels. + */ + return true; + default: + return false; + } +} + static int i10nm_get_ddr_munits(void) { struct pci_dev *mdev; void __iomem *mbase; unsigned long size; struct skx_dev *d; - int i, j = 0; + int i, lmc, j = 0; u32 reg, off; u64 base;
@@ -690,7 +726,7 @@ static int i10nm_get_ddr_munits(void) edac_dbg(2, "socket%d mmio base 0x%llx (reg 0x%x)\n", j++, base, reg);
- for (i = 0; i < res_cfg->ddr_imc_num; i++) { + for (lmc = 0, i = 0; i < res_cfg->ddr_imc_num; i++) { mdev = get_ddr_munit(d, i, &off, &size);
if (i == 0 && !mdev) { @@ -700,8 +736,6 @@ static int i10nm_get_ddr_munits(void) if (!mdev) continue;
- d->imc[i].mdev = mdev; - edac_dbg(2, "mc%d mmio base 0x%llx size 0x%lx (reg 0x%x)\n", i, base + off, size, reg);
@@ -712,7 +746,17 @@ static int i10nm_get_ddr_munits(void) return -ENODEV; }
- d->imc[i].mbase = mbase; + d->imc[lmc].mbase = mbase; + if (i10nm_imc_absent(&d->imc[lmc])) { + pci_dev_put(mdev); + iounmap(mbase); + d->imc[lmc].mbase = NULL; + edac_dbg(2, "Skip absent mc%d\n", i); + continue; + } else { + d->imc[lmc].mdev = mdev; + lmc++; + } } }