Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-09-25 Thread Linda Knippers
On 9/25/2014 9:56 AM, Rob Roschewsk wrote: > OK ... at first glance the system seems to be operating "normally" > except for the errors every 4 seconds ... that is it's attached to and > using the iscsi target ... > > Is this a problem in iommu OR the emulex driver?? Please make sure your BIOS

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-09-25 Thread Rob Roschewsk
OK ... at first glance the system seems to be operating "normally" except for the errors every 4 seconds ... that is it's attached to and using the iscsi target ... Is this a problem in iommu OR the emulex driver?? On Thu, Sep 25, 2014 at 2:36 AM, Yijing Wang wrote: > On 2014/9/25 5:56, Rob Ros

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-09-24 Thread Yijing Wang
On 2014/9/25 5:56, Rob Roschewsk wrote: > Hello All wonder if there has been any movement on this issue I'm > having a similar issue > > I'm running an HP dl380 gen 8 with an Emulex OneConnect 10Gb iSCSI > (14e4:164c) (rev 11) > also known as " Hewlett-Packard Company NC373i Integrat

RE: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-09-24 Thread Rob Roschewsk
Hello All wonder if there has been any movement on this issue I'm having a similar issue I'm running an HP dl380 gen 8 with an Emulex OneConnect 10Gb iSCSI (14e4:164c) (rev 11) also known as " Hewlett-Packard Company NC373i Integrated Multifunction Gigabit Server Adapter" 03:00.2 Mas

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-14 Thread Linda Knippers
On 8/11/2014 9:37 PM, Yijing Wang wrote: > On 2014/8/11 22:59, Linda Knippers wrote: >> On 8/11/2014 12:43 AM, Alex Williamson wrote: >>> On Mon, 2014-08-11 at 10:54 +0800, Yijing Wang wrote: We found some strange devices in HP C7000 and Huawei Server. These devices can not be enumerated

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-11 Thread Yijing Wang
On 2014/8/12 11:18, Jiang Liu wrote: > On 2014/8/12 9:37, Yijing Wang wrote: >> On 2014/8/11 22:59, Linda Knippers wrote: >>> On 8/11/2014 12:43 AM, Alex Williamson wrote: On Mon, 2014-08-11 at 10:54 +0800, Yijing Wang wrote: > We found some strange devices in HP C7000 and Huawei Server. T

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-11 Thread Jiang Liu
On 2014/8/12 9:37, Yijing Wang wrote: > On 2014/8/11 22:59, Linda Knippers wrote: >> On 8/11/2014 12:43 AM, Alex Williamson wrote: >>> On Mon, 2014-08-11 at 10:54 +0800, Yijing Wang wrote: We found some strange devices in HP C7000 and Huawei Server. These devices can not be enumerated by

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-11 Thread Jiang Liu
On 2014/8/12 9:37, Yijing Wang wrote: > On 2014/8/11 22:59, Linda Knippers wrote: >> On 8/11/2014 12:43 AM, Alex Williamson wrote: >>> On Mon, 2014-08-11 at 10:54 +0800, Yijing Wang wrote: We found some strange devices in HP C7000 and Huawei Server. These devices can not be enumerated b

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-11 Thread Linda Knippers
On 8/11/2014 12:43 AM, Alex Williamson wrote: > On Mon, 2014-08-11 at 10:54 +0800, Yijing Wang wrote: >> We found some strange devices in HP C7000 and Huawei Server. These devices >> can not be enumerated by OS, but they still did DMA read/write without OS >> management. Because iommu will not cre

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-11 Thread Yijing Wang
>> extern struct iommu_table_entry __iommu_table[], __iommu_table_end[]; >> >> @@ -146,6 +148,7 @@ void dma_generic_free_coherent(struct device *dev, >> size_t size, void *vaddr, >> */ >> static __init int iommu_setup(char *p) >> { >> +char *end; >> iommu_merge = 1; >> >> if

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-10 Thread Jiang Liu
Hi Yijing, These devices don't support Phantom Function (PhantFunc 0). Could you please help to provide full dmesg, /proc/iomem, and ACPI DMAR table? You may get DMAR table by: acpidump > acpi.bin acpixtract -a acpib.bin iasl -d DMAR.dat Thanks! Gerry On 2014/8/11 11:46, Yijing Wang wrote:

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-10 Thread Alex Williamson
On Mon, 2014-08-11 at 10:54 +0800, Yijing Wang wrote: > We found some strange devices in HP C7000 and Huawei Server. These devices > can not be enumerated by OS, but they still did DMA read/write without OS > management. Because iommu will not create the DMA mapping for these devices, > the DMA re

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-10 Thread Yijing Wang
On 2014/8/11 11:15, Jiang Liu wrote: > Hi Yijing, > Could you please help to provide detail information about the > PCI device, such as "lscpi -vvv"? It seems like that "Phantom > Functions" is enabled with these PCI devices. > Thanks! > Gerry My pleasure! Gerry, the lspci info is below: H

Re: [PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-10 Thread Jiang Liu
Hi Yijing, Could you please help to provide detail information about the PCI device, such as "lscpi -vvv"? It seems like that "Phantom Functions" is enabled with these PCI devices. Thanks! Gerry On 2014/8/11 10:54, Yijing Wang wrote: > We found some strange devices in HP C7000 and Huawei S

[PATCH] iommu/vt-d: Fix broken device issue when using iommu=pt

2014-08-10 Thread Yijing Wang
We found some strange devices in HP C7000 and Huawei Server. These devices can not be enumerated by OS, but they still did DMA read/write without OS management. Because iommu will not create the DMA mapping for these devices, the DMA read/write will be blocked by iommu hardware. Eg. \-[:00]-