mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Narayana Murty N <nnmlinux@linux.ibm.com>
To: Sourabh Jain <sourabhjain@linux.ibm.com>,
	Mahesh J Salgaonkar <mahesh@linux.ibm.com>,
	Madhavan Srinivasan <maddy@linux.ibm.com>,
	"Oliver O'Halloran" <oohall@gmail.com>,
	Timothy Pearson <tpearson@raptorengineering.com>,
	Alex Williamson <alex@shazbot.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>,
	Nicholas Piggin <npiggin@gmail.com>,
	Christophe Leroy <chleroy@kernel.org>,
	Ritesh Harjani <ritesh.list@gmail.com>,
	Shrikanth Hegde <sshegde@linux.ibm.com>,
	linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org,
	kvm@vger.kernel.org, sbhat@linux.ibm.com, harshpb@linux.ibm.com
Subject: Re: [PATCH v6 4/5] vfio/spapr_tce: Normalize EEH IOA error injection addresses
Date: Fri, 20 Nov 2026 13:06:18 +0530	[thread overview]
Message-ID: <1b52c142-116c-4a9e-af19-578a9a71e3f3@linux.ibm.com> (raw)
In-Reply-To: <518b5813-ab4e-413e-9820-556888216090@linux.ibm.com>



On 19/11/26 1:05 AM, Sourabh Jain wrote:
>   I'm finding this commit message a bit difficult to follow.
> Could you please rework it to make the changes and their purpose clearer?
> 
Agreed. I'll rework the commit message to focus on why address
translation is needed in the VFIO EEH path.
> On 23/10/26 13:06, Narayana Murty N wrote:
>> VFIO_EEH_PE_INJECT_ERR receives a target address from userspace.  For
>> userspace such as QEMU, this address may be derived from the host Linux
>> BAR resource.  The platform EEH backends, however, expect the PCI/IOA
>> bus address used by firmware error-injection interfaces.
>>
>> Normalize IOA/MMIO error-injection addresses in the sPAPR VFIO EEH
>> ioctl path before dispatching to eeh_pe_inject_err().  If the supplied
>> address is already a PCI bus BAR address it is left unchanged.  If it is
>> a Linux resource address, translate it to the corresponding PCI bus
>> address using the BAR-relative offset.
>>
>> Normalization applies only to EEH_ERR_TYPE_32 and EEH_ERR_TYPE_64 error
>> types whose error function targets a memory or IO BAR (load/store memory
>> and load/store IO functions, func 0-3 and 6-9).  Config-space functions
>> carry PCI config register offsets and DMA functions carry IOVA/TCE
>> addresses; these address domains are not PCI BAR addresses and must not
>> be translated.  Applying BAR normalization to them would incorrectly
>> return -EINVAL, blocking injection entirely for those function codes.
>>
>> An address and mask pair of zero represents a no-address injection
>> (e.g. a function-level error with no specific IOA target) and is passed
>> through without BAR translation.
>>
>> Only standard BARs (BAR0..BAR5) are searched; ROM and bridge resources
>> are not IOA bus-error targets.  Unassigned or disabled BARs are skipped:
>> pci_resource_len() returns 0 for zero-initialized BARs, and the
>> IORESOURCE_UNSET and IORESOURCE_DISABLED flags are checked explicitly to
>> avoid false matches against unassigned resources.  Hold
>> pci_rescan_remove_lock while traversing PE devices and their BAR
>> resources to prevent PCI device removal from racing the lookup.
>>
>> A pr_debug() is emitted if the address cannot be matched to any BAR, to
>> aid debugging without producing noise in the normal path.
> Which address are you referring to?
> 
The target address supplied by userspace through
VFIO_EEH_PE_INJECT_ERR, which may be a Linux resource address or
a PCI bus address.

I'll make this explicit in the commit message.
> 
>>
>> Keep the helper local to VFIO so the address semantics of other
>> in-kernel EEH callers remain unchanged.
> 
> Which helper are you referring to?
> 
I was referring to vfio_spapr_eeh_normalize_addr().

The intention is to keep the resource-to-bus address translation
local to the VFIO path without changing the address semantics of
other EEH callers.
> 
>>
>> Signed-off-by: Narayana Murty N <nnmlinux@linux.ibm.com>
>> ---
>>   drivers/vfio/vfio_iommu_spapr_tce.c | 140 ++++++++++++++++++++++++++++
>>   1 file changed, 140 insertions(+)
>>
>> diff --git a/drivers/vfio/vfio_iommu_spapr_tce.c b/drivers/vfio/ 
>> vfio_iommu_spapr_tce.c
>> index 1c0eec228cc6..d831adaa69fb 100644
>> --- a/drivers/vfio/vfio_iommu_spapr_tce.c
>> +++ b/drivers/vfio/vfio_iommu_spapr_tce.c
>> @@ -774,6 +774,131 @@ static long 
>> tce_iommu_create_default_window(struct tce_container *container)
>>       return ret;
>>   }
>> +/*
>> + * eeh_func_needs_bar_addr - return true for error functions whose addr
>> + * argument is a PCI MMIO or IO BAR address.
>> + *
>> + * Memory and IO load/store functions target specific BAR addresses.
>> + * Config-space functions carry PCI config register offsets, and DMA
>> + * functions carry IOVA/TCE addresses; neither requires BAR translation.
>> + */
>> +static bool eeh_func_needs_bar_addr(int func)
>> +{
>> +    switch (func) {
>> +    case EEH_ERR_FUNC_LD_MEM_ADDR:
>> +    case EEH_ERR_FUNC_LD_MEM_DATA:
>> +    case EEH_ERR_FUNC_LD_IO_ADDR:
>> +    case EEH_ERR_FUNC_LD_IO_DATA:
>> +    case EEH_ERR_FUNC_ST_MEM_ADDR:
>> +    case EEH_ERR_FUNC_ST_MEM_DATA:
>> +    case EEH_ERR_FUNC_ST_IO_ADDR:
>> +    case EEH_ERR_FUNC_ST_IO_DATA:
>> +        return true;
>> +    default:
>> +        return false;
>> +    }
>> +}
>> +
>> +static int vfio_spapr_eeh_normalize_addr(struct eeh_pe *pe,
>> +                     unsigned long addr,
>> +                     unsigned long mask,
> Do we really need mask argument?
> 

The mask argument is currently used only to identify the no-address
injection case where both addr and mask are zero.

I'll move that check to the ioctl caller and remove the mask
argument from the normalization helper.

>> +                     unsigned long *normalized)
> 
> Too much in a single function. Can you split this function as suggested
> for validate_addr_mask_in_pe() in 2/5.
> 
Agreed. I'll separate the per-device BAR matching and address.

> - Sourabh Jain
> 
>> +{
>> +    struct pci_bus_region region;
>> +    struct eeh_dev *edev, *tmp;
>> +    struct pci_dev *pdev;
>> +    struct resource *res;
>> +    resource_size_t pci_start, pci_len;
>> +    resource_size_t res_start, res_len;
>> +    resource_size_t offset;
>> +    int bar, rc = -EINVAL;
>> +
>> +    if (!pe || !normalized)
>> +        return -EINVAL;
>> +
>> +    /*
>> +     * addr == 0 and mask == 0 indicates a no-address injection (e.g.
>> +     * function-level error with no specific IOA target).  No BAR
>> +     * translation needed.
>> +     */
>> +    if (!addr && !mask) {
>> +        *normalized = addr;
>> +        return 0;
>> +    }
>> +
>> +    /*
>> +     * Walk standard BARs (BAR0..BAR5) of all devices in the PE.
>> +     * Hold pci_rescan_remove_lock to protect pdev lifetime during the
>> +     * traversal: pci_stop_and_remove_bus_device() requires this lock.
>> +     * ROM and bridge resources are not IOA bus-error targets, so the
>> +     * search is limited to PCI_STD_NUM_BARS.
>> +     */
>> +    pci_lock_rescan_remove();
>> +    eeh_pe_for_each_dev(pe, edev, tmp) {
>> +        pdev = eeh_dev_to_pci_dev(edev);
>> +        if (!pdev)
>> +            continue;
>> +
>> +        for (bar = 0; bar < PCI_STD_NUM_BARS; bar++) {
>> +            res = &pdev->resource[bar];
>> +
>> +            /*
>> +             * Skip unassigned or disabled BARs.  On pseries,
>> +             * pcibios_fixup_resources() zero-bases unassigned BARs
>> +             * (start=0, end=size-1) and sets IORESOURCE_UNSET.
>> +             * Use pci_resource_len() which returns 0 when end==0.
>> +             */
>> +            if (!pci_resource_len(pdev, bar))
>> +                continue;
>> +
>> +            if (!(res->flags & (IORESOURCE_MEM | IORESOURCE_IO)))
>> +                continue;
>> +
>> +            if (res->flags & (IORESOURCE_UNSET | IORESOURCE_DISABLED))
>> +                continue;
>> +
>> +            pcibios_resource_to_bus(pdev->bus, &region, res);
>> +
>> +            pci_start = region.start;
>> +            pci_len = pci_resource_len(pdev, bar);
>> +
>> +            /*
>> +             * Case 1: userspace already supplied PCI/IOA
>> +             * bus address.
>> +             */
>> +            if ((resource_size_t)addr >= pci_start &&
>> +                ((resource_size_t)addr - pci_start) < pci_len) {
>> +                *normalized = addr;
>> +                rc = 0;
>> +                goto done;
>> +            }
>> +
>> +            /*
>> +             * Case 2: userspace supplied Linux resource/CPU
>> +             * address.  Convert it back to PCI/IOA bus address
>> +             * before calling the platform EEH backend.
>> +             */
>> +            res_start = res->start;
>> +            res_len = pci_resource_len(pdev, bar);
>> +
>> +            if ((resource_size_t)addr >= res_start &&
>> +                ((resource_size_t)addr - res_start) < res_len) {
>> +                offset = (resource_size_t)addr - res_start;
>> +                *normalized = region.start + offset;
>> +                rc = 0;
>> +                goto done;
>> +            }
>> +        }
>> +    }
>> +done:
>> +    pci_unlock_rescan_remove();
>> +
>> +    if (rc)
>> +        pr_debug("vfio EEH: addr=0x%lx not found in any PCI bus BAR 
>> of PE\n",
>> +             addr);
>> +    return rc;
>> +}
>> +
>>   static long vfio_spapr_ioctl_eeh_pe_op(struct iommu_group *group,
>>                          unsigned long arg)
>>   {
>> @@ -818,6 +943,21 @@ static long vfio_spapr_ioctl_eeh_pe_op(struct 
>> iommu_group *group,
>>           if (copy_from_user(&op, (void __user *)arg, minsz))
>>               return -EFAULT;
>> +        if ((op.err.type == EEH_ERR_TYPE_32 ||
>> +             op.err.type == EEH_ERR_TYPE_64) &&
>> +            eeh_func_needs_bar_addr(op.err.func)) {
>> +            unsigned long normalized;
>> +            int ret;
>> +
>> +            ret = vfio_spapr_eeh_normalize_addr(pe, op.err.addr,
>> +                                op.err.mask,
>> +                                &normalized);
>> +            if (ret)
>> +                return ret;
>> +
>> +            op.err.addr = normalized;
>> +        }
>> +
>>           return eeh_pe_inject_err(pe, op.err.type, op.err.func,
>>                        op.err.addr, op.err.mask);
>>       default:

Thanks,
Narayana Murty.
> 


  reply	other threads:[~2026-11-20  7:36 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-23  7:36 [PATCH v6 0/5] powerpc: Add RTAS based error injection support for pSeries Narayana Murty N
2026-10-23  7:36 ` [PATCH v6 1/5] powerpc/rtas: Handle ibm,open-errinjct return format Narayana Murty N
2026-11-18 11:13   ` Sourabh Jain
2026-11-20  4:36     ` Narayana Murty N
2026-10-23  7:36 ` [PATCH v6 2/5] powerpc/pseries/eeh: Add RTAS error injection helpers Narayana Murty N
2026-11-18 16:43   ` Sourabh Jain
2026-11-20  7:19     ` Narayana Murty N
2026-10-23  7:36 ` [PATCH v6 3/5] powerpc/pseries/eeh: Implement RTAS-based EEH error injection Narayana Murty N
2026-11-18 19:14   ` Sourabh Jain
2026-11-20  7:32     ` Narayana Murty N
2026-10-23  7:36 ` [PATCH v6 4/5] vfio/spapr_tce: Normalize EEH IOA error injection addresses Narayana Murty N
2026-11-18 19:35   ` Sourabh Jain
2026-11-20  7:36     ` Narayana Murty N [this message]
2026-10-23  7:36 ` [PATCH v6 5/5] powerpc/powernv/eeh: Map generic EEH error types to OPAL Narayana Murty N
2026-11-18 19:41   ` Sourabh Jain
2026-11-20  7:39     ` Narayana Murty N
2026-11-10 19:21 ` [PATCH v6 0/5] powerpc: Add RTAS based error injection support for pSeries tasmiya
2026-11-20  7:40   ` Narayana Murty N

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1b52c142-116c-4a9e-af19-578a9a71e3f3@linux.ibm.com \
    --to=nnmlinux@linux.ibm.com \
    --cc=alex@shazbot.org \
    --cc=chleroy@kernel.org \
    --cc=harshpb@linux.ibm.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linuxppc-dev@lists.ozlabs.org \
    --cc=maddy@linux.ibm.com \
    --cc=mahesh@linux.ibm.com \
    --cc=mpe@ellerman.id.au \
    --cc=npiggin@gmail.com \
    --cc=oohall@gmail.com \
    --cc=ritesh.list@gmail.com \
    --cc=sbhat@linux.ibm.com \
    --cc=sourabhjain@linux.ibm.com \
    --cc=sshegde@linux.ibm.com \
    --cc=tpearson@raptorengineering.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®