[PATCH] powerpc/powernv: fix endianness problems in EEH
Gavin Shan
gwshan at linux.vnet.ibm.com
Tue Jun 10 10:28:16 EST 2014
On Mon, Jun 09, 2014 at 04:58:51PM +0800, Guo Chao wrote:
>EEH information fetched from OPAL need fix before using in LE environment.
>To be included in sparse's endian check, declare them as __beXX and
>access them by accessors.
>
>Cc: Gavin Shan <gwshan at linux.vnet.ibm.com>
>
>Signed-off-by: Guo Chao <yan at linux.vnet.ibm.com>
Acked-by: Gavin Shan <gwshan at linux.vnet.ibm.com>
It looks good to me., but there are more diag-data needed to be
coverted: P7IOC diag-data and P7IOC PHB diag-data.
Thanks,
Gavin
>---
> arch/powerpc/include/asm/opal.h | 102 +++++++++++++++---------------
> arch/powerpc/platforms/powernv/eeh-ioda.c | 36 ++++++-----
> arch/powerpc/platforms/powernv/pci.c | 81 +++++++++++++++---------
> 3 files changed, 120 insertions(+), 99 deletions(-)
>
>diff --git a/arch/powerpc/include/asm/opal.h b/arch/powerpc/include/asm/opal.h
>index 4ccd1d4..a4c0acc 100644
>--- a/arch/powerpc/include/asm/opal.h
>+++ b/arch/powerpc/include/asm/opal.h
>@@ -599,9 +599,9 @@ enum {
> };
>
> struct OpalIoPhbErrorCommon {
>- uint32_t version;
>- uint32_t ioType;
>- uint32_t len;
>+ __be32 version;
>+ __be32 ioType;
>+ __be32 len;
> };
>
> struct OpalIoP7IOCPhbErrorData {
>@@ -671,64 +671,64 @@ enum {
> struct OpalIoPhb3ErrorData {
> struct OpalIoPhbErrorCommon common;
>
>- uint32_t brdgCtl;
>+ __be32 brdgCtl;
>
> /* PHB3 UTL regs */
>- uint32_t portStatusReg;
>- uint32_t rootCmplxStatus;
>- uint32_t busAgentStatus;
>+ __be32 portStatusReg;
>+ __be32 rootCmplxStatus;
>+ __be32 busAgentStatus;
>
> /* PHB3 cfg regs */
>- uint32_t deviceStatus;
>- uint32_t slotStatus;
>- uint32_t linkStatus;
>- uint32_t devCmdStatus;
>- uint32_t devSecStatus;
>+ __be32 deviceStatus;
>+ __be32 slotStatus;
>+ __be32 linkStatus;
>+ __be32 devCmdStatus;
>+ __be32 devSecStatus;
>
> /* cfg AER regs */
>- uint32_t rootErrorStatus;
>- uint32_t uncorrErrorStatus;
>- uint32_t corrErrorStatus;
>- uint32_t tlpHdr1;
>- uint32_t tlpHdr2;
>- uint32_t tlpHdr3;
>- uint32_t tlpHdr4;
>- uint32_t sourceId;
>+ __be32 rootErrorStatus;
>+ __be32 uncorrErrorStatus;
>+ __be32 corrErrorStatus;
>+ __be32 tlpHdr1;
>+ __be32 tlpHdr2;
>+ __be32 tlpHdr3;
>+ __be32 tlpHdr4;
>+ __be32 sourceId;
>
>- uint32_t rsv3;
>+ __be32 rsv3;
>
> /* Record data about the call to allocate a buffer */
>- uint64_t errorClass;
>- uint64_t correlator;
>+ __be64 errorClass;
>+ __be64 correlator;
>
>- uint64_t nFir; /* 000 */
>- uint64_t nFirMask; /* 003 */
>- uint64_t nFirWOF; /* 008 */
>+ __be64 nFir; /* 000 */
>+ __be64 nFirMask; /* 003 */
>+ __be64 nFirWOF; /* 008 */
>
> /* PHB3 MMIO Error Regs */
>- uint64_t phbPlssr; /* 120 */
>- uint64_t phbCsr; /* 110 */
>- uint64_t lemFir; /* C00 */
>- uint64_t lemErrorMask; /* C18 */
>- uint64_t lemWOF; /* C40 */
>- uint64_t phbErrorStatus; /* C80 */
>- uint64_t phbFirstErrorStatus; /* C88 */
>- uint64_t phbErrorLog0; /* CC0 */
>- uint64_t phbErrorLog1; /* CC8 */
>- uint64_t mmioErrorStatus; /* D00 */
>- uint64_t mmioFirstErrorStatus; /* D08 */
>- uint64_t mmioErrorLog0; /* D40 */
>- uint64_t mmioErrorLog1; /* D48 */
>- uint64_t dma0ErrorStatus; /* D80 */
>- uint64_t dma0FirstErrorStatus; /* D88 */
>- uint64_t dma0ErrorLog0; /* DC0 */
>- uint64_t dma0ErrorLog1; /* DC8 */
>- uint64_t dma1ErrorStatus; /* E00 */
>- uint64_t dma1FirstErrorStatus; /* E08 */
>- uint64_t dma1ErrorLog0; /* E40 */
>- uint64_t dma1ErrorLog1; /* E48 */
>- uint64_t pestA[OPAL_PHB3_NUM_PEST_REGS];
>- uint64_t pestB[OPAL_PHB3_NUM_PEST_REGS];
>+ __be64 phbPlssr; /* 120 */
>+ __be64 phbCsr; /* 110 */
>+ __be64 lemFir; /* C00 */
>+ __be64 lemErrorMask; /* C18 */
>+ __be64 lemWOF; /* C40 */
>+ __be64 phbErrorStatus; /* C80 */
>+ __be64 phbFirstErrorStatus; /* C88 */
>+ __be64 phbErrorLog0; /* CC0 */
>+ __be64 phbErrorLog1; /* CC8 */
>+ __be64 mmioErrorStatus; /* D00 */
>+ __be64 mmioFirstErrorStatus; /* D08 */
>+ __be64 mmioErrorLog0; /* D40 */
>+ __be64 mmioErrorLog1; /* D48 */
>+ __be64 dma0ErrorStatus; /* D80 */
>+ __be64 dma0FirstErrorStatus; /* D88 */
>+ __be64 dma0ErrorLog0; /* DC0 */
>+ __be64 dma0ErrorLog1; /* DC8 */
>+ __be64 dma1ErrorStatus; /* E00 */
>+ __be64 dma1FirstErrorStatus; /* E08 */
>+ __be64 dma1ErrorLog0; /* E40 */
>+ __be64 dma1ErrorLog1; /* E48 */
>+ __be64 pestA[OPAL_PHB3_NUM_PEST_REGS];
>+ __be64 pestB[OPAL_PHB3_NUM_PEST_REGS];
> };
>
> typedef struct oppanel_line {
>@@ -851,8 +851,8 @@ int64_t opal_pci_mask_pe_error(uint64_t phb_id, uint16_t pe_number, uint8_t erro
> int64_t opal_set_slot_led_status(uint64_t phb_id, uint64_t slot_id, uint8_t led_type, uint8_t led_action);
> int64_t opal_get_epow_status(__be64 *status);
> int64_t opal_set_system_attention_led(uint8_t led_action);
>-int64_t opal_pci_next_error(uint64_t phb_id, uint64_t *first_frozen_pe,
>- uint16_t *pci_error_type, uint16_t *severity);
>+int64_t opal_pci_next_error(uint64_t phb_id, __be64 *first_frozen_pe,
>+ __be16 *pci_error_type, __be16 *severity);
> int64_t opal_pci_poll(uint64_t phb_id);
> int64_t opal_return_cpu(void);
> int64_t opal_reinit_cpus(uint64_t flags);
>diff --git a/arch/powerpc/platforms/powernv/eeh-ioda.c b/arch/powerpc/platforms/powernv/eeh-ioda.c
>index 753f08e..e0d6a3a 100644
>--- a/arch/powerpc/platforms/powernv/eeh-ioda.c
>+++ b/arch/powerpc/platforms/powernv/eeh-ioda.c
>@@ -267,7 +267,7 @@ static int ioda_eeh_get_state(struct eeh_pe *pe)
> {
> s64 ret = 0;
> u8 fstate;
>- u16 pcierr;
>+ __be16 pcierr;
> u32 pe_no;
> int result;
> struct pci_controller *hose = pe->phb;
>@@ -316,7 +316,7 @@ static int ioda_eeh_get_state(struct eeh_pe *pe)
> result = 0;
> result &= ~EEH_STATE_RESET_ACTIVE;
>
>- if (pcierr != OPAL_EEH_PHB_ERROR) {
>+ if (be16_to_cpu(pcierr) != OPAL_EEH_PHB_ERROR) {
> result |= EEH_STATE_MMIO_ACTIVE;
> result |= EEH_STATE_DMA_ACTIVE;
> result |= EEH_STATE_MMIO_ENABLED;
>@@ -706,8 +706,8 @@ static int ioda_eeh_next_error(struct eeh_pe **pe)
> struct pci_controller *hose;
> struct pnv_phb *phb;
> struct eeh_pe *phb_pe;
>- u64 frozen_pe_no;
>- u16 err_type, severity;
>+ __be64 frozen_pe_no;
>+ __be16 err_type, severity;
> long rc;
> int ret = EEH_NEXT_ERR_NONE;
>
>@@ -742,8 +742,8 @@ static int ioda_eeh_next_error(struct eeh_pe **pe)
> }
>
> /* If the PHB doesn't have error, stop processing */
>- if (err_type == OPAL_EEH_NO_ERROR ||
>- severity == OPAL_EEH_SEV_NO_ERROR) {
>+ if (be16_to_cpu(err_type) == OPAL_EEH_NO_ERROR ||
>+ be16_to_cpu(severity) == OPAL_EEH_SEV_NO_ERROR) {
> pr_devel("%s: No error found on PHB#%x\n",
> __func__, hose->global_number);
> continue;
>@@ -755,14 +755,14 @@ static int ioda_eeh_next_error(struct eeh_pe **pe)
> * specific PHB.
> */
> pr_devel("%s: Error (%d, %d, %llu) on PHB#%x\n",
>- __func__, err_type, severity,
>- frozen_pe_no, hose->global_number);
>- switch (err_type) {
>+ __func__, be16_to_cpu(err_type), be16_to_cpu(severity),
>+ be64_to_cpu(frozen_pe_no), hose->global_number);
>+ switch (be16_to_cpu(err_type)) {
> case OPAL_EEH_IOC_ERROR:
>- if (severity == OPAL_EEH_SEV_IOC_DEAD) {
>+ if (be16_to_cpu(severity) == OPAL_EEH_SEV_IOC_DEAD) {
> pr_err("EEH: dead IOC detected\n");
> ret = EEH_NEXT_ERR_DEAD_IOC;
>- } else if (severity == OPAL_EEH_SEV_INF) {
>+ } else if (be16_to_cpu(severity) == OPAL_EEH_SEV_INF) {
> pr_info("EEH: IOC informative error "
> "detected\n");
> ioda_eeh_hub_diag(hose);
>@@ -771,17 +771,18 @@ static int ioda_eeh_next_error(struct eeh_pe **pe)
>
> break;
> case OPAL_EEH_PHB_ERROR:
>- if (severity == OPAL_EEH_SEV_PHB_DEAD) {
>+ if (be16_to_cpu(severity) == OPAL_EEH_SEV_PHB_DEAD) {
> *pe = phb_pe;
> pr_err("EEH: dead PHB#%x detected\n",
> hose->global_number);
> ret = EEH_NEXT_ERR_DEAD_PHB;
>- } else if (severity == OPAL_EEH_SEV_PHB_FENCED) {
>+ } else if (be16_to_cpu(severity) ==
>+ OPAL_EEH_SEV_PHB_FENCED) {
> *pe = phb_pe;
> pr_err("EEH: fenced PHB#%x detected\n",
> hose->global_number);
> ret = EEH_NEXT_ERR_FENCED_PHB;
>- } else if (severity == OPAL_EEH_SEV_INF) {
>+ } else if (be16_to_cpu(severity) == OPAL_EEH_SEV_INF) {
> pr_info("EEH: PHB#%x informative error "
> "detected\n",
> hose->global_number);
>@@ -801,12 +802,13 @@ static int ioda_eeh_next_error(struct eeh_pe **pe)
> * progress with recovery. We needn't report
> * it again.
> */
>- if (ioda_eeh_get_pe(hose, frozen_pe_no, pe)) {
>+ if (ioda_eeh_get_pe(hose,
>+ be64_to_cpu(frozen_pe_no), pe)) {
> *pe = phb_pe;
> pr_err("EEH: Escalated fenced PHB#%x "
> "detected for PE#%llx\n",
> hose->global_number,
>- frozen_pe_no);
>+ be64_to_cpu(frozen_pe_no));
> ret = EEH_NEXT_ERR_FENCED_PHB;
> } else if ((*pe)->state & EEH_PE_ISOLATED) {
> ret = EEH_NEXT_ERR_NONE;
>@@ -819,7 +821,7 @@ static int ioda_eeh_next_error(struct eeh_pe **pe)
> break;
> default:
> pr_warn("%s: Unexpected error type %d\n",
>- __func__, err_type);
>+ __func__, be16_to_cpu(err_type));
> }
>
> /*
>diff --git a/arch/powerpc/platforms/powernv/pci.c b/arch/powerpc/platforms/powernv/pci.c
>index eefbfcc..f91a4e5 100644
>--- a/arch/powerpc/platforms/powernv/pci.c
>+++ b/arch/powerpc/platforms/powernv/pci.c
>@@ -206,72 +206,91 @@ static void pnv_pci_dump_phb3_diag_data(struct pci_controller *hose,
>
> data = (struct OpalIoPhb3ErrorData*)common;
> pr_info("PHB3 PHB#%d Diag-data (Version: %d)\n",
>- hose->global_number, common->version);
>+ hose->global_number, be32_to_cpu(common->version));
> if (data->brdgCtl)
> pr_info("brdgCtl: %08x\n",
>- data->brdgCtl);
>+ be32_to_cpu(data->brdgCtl));
> if (data->portStatusReg || data->rootCmplxStatus ||
> data->busAgentStatus)
> pr_info("UtlSts: %08x %08x %08x\n",
>- data->portStatusReg, data->rootCmplxStatus,
>- data->busAgentStatus);
>+ be32_to_cpu(data->portStatusReg),
>+ be32_to_cpu(data->rootCmplxStatus),
>+ be32_to_cpu(data->busAgentStatus));
> if (data->deviceStatus || data->slotStatus ||
> data->linkStatus || data->devCmdStatus ||
> data->devSecStatus)
> pr_info("RootSts: %08x %08x %08x %08x %08x\n",
>- data->deviceStatus, data->slotStatus,
>- data->linkStatus, data->devCmdStatus,
>- data->devSecStatus);
>+ be32_to_cpu(data->deviceStatus),
>+ be32_to_cpu(data->slotStatus),
>+ be32_to_cpu(data->linkStatus),
>+ be32_to_cpu(data->devCmdStatus),
>+ be32_to_cpu(data->devSecStatus));
> if (data->rootErrorStatus || data->uncorrErrorStatus ||
> data->corrErrorStatus)
> pr_info("RootErrSts: %08x %08x %08x\n",
>- data->rootErrorStatus, data->uncorrErrorStatus,
>- data->corrErrorStatus);
>+ be32_to_cpu(data->rootErrorStatus),
>+ be32_to_cpu(data->uncorrErrorStatus),
>+ be32_to_cpu(data->corrErrorStatus));
> if (data->tlpHdr1 || data->tlpHdr2 ||
> data->tlpHdr3 || data->tlpHdr4)
> pr_info("RootErrLog: %08x %08x %08x %08x\n",
>- data->tlpHdr1, data->tlpHdr2,
>- data->tlpHdr3, data->tlpHdr4);
>+ be32_to_cpu(data->tlpHdr1),
>+ be32_to_cpu(data->tlpHdr2),
>+ be32_to_cpu(data->tlpHdr3),
>+ be32_to_cpu(data->tlpHdr4));
> if (data->sourceId || data->errorClass ||
> data->correlator)
> pr_info("RootErrLog1: %08x %016llx %016llx\n",
>- data->sourceId, data->errorClass,
>- data->correlator);
>+ be32_to_cpu(data->sourceId),
>+ be64_to_cpu(data->errorClass),
>+ be64_to_cpu(data->correlator));
> if (data->nFir)
> pr_info("nFir: %016llx %016llx %016llx\n",
>- data->nFir, data->nFirMask,
>- data->nFirWOF);
>+ be64_to_cpu(data->nFir),
>+ be64_to_cpu(data->nFirMask),
>+ be64_to_cpu(data->nFirWOF));
> if (data->phbPlssr || data->phbCsr)
> pr_info("PhbSts: %016llx %016llx\n",
>- data->phbPlssr, data->phbCsr);
>+ be64_to_cpu(data->phbPlssr),
>+ be64_to_cpu(data->phbCsr));
> if (data->lemFir)
> pr_info("Lem: %016llx %016llx %016llx\n",
>- data->lemFir, data->lemErrorMask,
>- data->lemWOF);
>+ be64_to_cpu(data->lemFir),
>+ be64_to_cpu(data->lemErrorMask),
>+ be64_to_cpu(data->lemWOF));
> if (data->phbErrorStatus)
> pr_info("PhbErr: %016llx %016llx %016llx %016llx\n",
>- data->phbErrorStatus, data->phbFirstErrorStatus,
>- data->phbErrorLog0, data->phbErrorLog1);
>+ be64_to_cpu(data->phbErrorStatus),
>+ be64_to_cpu(data->phbFirstErrorStatus),
>+ be64_to_cpu(data->phbErrorLog0),
>+ be64_to_cpu(data->phbErrorLog1));
> if (data->mmioErrorStatus)
> pr_info("OutErr: %016llx %016llx %016llx %016llx\n",
>- data->mmioErrorStatus, data->mmioFirstErrorStatus,
>- data->mmioErrorLog0, data->mmioErrorLog1);
>+ be64_to_cpu(data->mmioErrorStatus),
>+ be64_to_cpu(data->mmioFirstErrorStatus),
>+ be64_to_cpu(data->mmioErrorLog0),
>+ be64_to_cpu(data->mmioErrorLog1));
> if (data->dma0ErrorStatus)
> pr_info("InAErr: %016llx %016llx %016llx %016llx\n",
>- data->dma0ErrorStatus, data->dma0FirstErrorStatus,
>- data->dma0ErrorLog0, data->dma0ErrorLog1);
>+ be64_to_cpu(data->dma0ErrorStatus),
>+ be64_to_cpu(data->dma0FirstErrorStatus),
>+ be64_to_cpu(data->dma0ErrorLog0),
>+ be64_to_cpu(data->dma0ErrorLog1));
> if (data->dma1ErrorStatus)
> pr_info("InBErr: %016llx %016llx %016llx %016llx\n",
>- data->dma1ErrorStatus, data->dma1FirstErrorStatus,
>- data->dma1ErrorLog0, data->dma1ErrorLog1);
>+ be64_to_cpu(data->dma1ErrorStatus),
>+ be64_to_cpu(data->dma1FirstErrorStatus),
>+ be64_to_cpu(data->dma1ErrorLog0),
>+ be64_to_cpu(data->dma1ErrorLog1));
>
> for (i = 0; i < OPAL_PHB3_NUM_PEST_REGS; i++) {
>- if ((data->pestA[i] >> 63) == 0 &&
>- (data->pestB[i] >> 63) == 0)
>+ if ((be64_to_cpu(data->pestA[i]) >> 63) == 0 &&
>+ (be64_to_cpu(data->pestB[i]) >> 63) == 0)
> continue;
>
> pr_info("PE[%3d] A/B: %016llx %016llx\n",
>- i, data->pestA[i], data->pestB[i]);
>+ i, be64_to_cpu(data->pestA[i]),
>+ be64_to_cpu(data->pestB[i]));
> }
> }
>
>@@ -284,7 +303,7 @@ void pnv_pci_dump_phb_diag_data(struct pci_controller *hose,
> return;
>
> common = (struct OpalIoPhbErrorCommon *)log_buff;
>- switch (common->ioType) {
>+ switch (be32_to_cpu(common->ioType)) {
> case OPAL_PHB_ERROR_DATA_TYPE_P7IOC:
> pnv_pci_dump_p7ioc_diag_data(hose, common);
> break;
>@@ -293,7 +312,7 @@ void pnv_pci_dump_phb_diag_data(struct pci_controller *hose,
> break;
> default:
> pr_warn("%s: Unrecognized ioType %d\n",
>- __func__, common->ioType);
>+ __func__, be32_to_cpu(common->ioType));
> }
> }
>
>--
>1.9.1
>
More information about the Linuxppc-dev
mailing list