From: Wei Wang <wei.wang2@amd.com>
To: Santosh Jodh <santosh.jodh@citrix.com>
Cc: xen-devel@lists.xensource.com, tim@xen.org, xiantao.zhang@intel.com
Subject: Re: [PATCH] dump_p2m_table: For IOMMU
Date: Fri, 10 Aug 2012 14:31:06 +0200 [thread overview]
Message-ID: <5024FF0A.6090006@amd.com> (raw)
In-Reply-To: <f687c55802629c72405d.1344562989@REDBLD-XS.ad.xensource.com>
Hi Santosh,
On 08/10/2012 03:43 AM, Santosh Jodh wrote:
> New key handler 'o' to dump the IOMMU p2m table for each domain.
> Skips dumping table for domain0.
> Intel and AMD specific iommu_ops handler for dumping p2m table.
>
> Signed-off-by: Santosh Jodh<santosh.jodh@citrix.com>
>
> diff -r 472fc515a463 -r f687c5580262 xen/drivers/passthrough/amd/pci_amd_iommu.c
> --- a/xen/drivers/passthrough/amd/pci_amd_iommu.c Tue Aug 07 18:37:31 2012 +0100
> +++ b/xen/drivers/passthrough/amd/pci_amd_iommu.c Thu Aug 09 18:31:32 2012 -0700
> @@ -22,6 +22,7 @@
> #include<xen/pci.h>
> #include<xen/pci_regs.h>
> #include<xen/paging.h>
> +#include<xen/softirq.h>
> #include<asm/hvm/iommu.h>
> #include<asm/amd-iommu.h>
> #include<asm/hvm/svm/amd-iommu-proto.h>
> @@ -512,6 +513,81 @@ static int amd_iommu_group_id(u16 seg, u
>
> #include<asm/io_apic.h>
>
> +static void amd_dump_p2m_table_level(struct page_info* pg, int level,
> + paddr_t gpa, int *indent)
> +{
> + paddr_t address;
> + void *table_vaddr, *pde;
> + paddr_t next_table_maddr;
> + int index, next_level, present;
> + u32 *entry;
> +
> + if ( level<= 1 )
> + return;
So, the l1 page table is not printed, is it by design?
> + table_vaddr = __map_domain_page(pg);
> + if ( table_vaddr == NULL )
> + {
> + printk("Failed to map IOMMU domain page %"PRIpaddr"\n",
> + page_to_maddr(pg));
> + return;
> + }
> +
> + for ( index = 0; index< PTE_PER_TABLE_SIZE; index++ )
> + {
> + if ( !(index % 2) )
> + process_pending_softirqs();
> +
> + pde = table_vaddr + (index * IOMMU_PAGE_TABLE_ENTRY_SIZE);
> + next_table_maddr = amd_iommu_get_next_table_from_pte(pde);
> + entry = (u32*)pde;
> +
> + next_level = get_field_from_reg_u32(entry[0],
> + IOMMU_PDE_NEXT_LEVEL_MASK,
> + IOMMU_PDE_NEXT_LEVEL_SHIFT);
> +
> + present = get_field_from_reg_u32(entry[0],
> + IOMMU_PDE_PRESENT_MASK,
> + IOMMU_PDE_PRESENT_SHIFT);
> +
> + address = gpa + amd_offset_level_address(index, level);
> + if ( present )
> + {
> + int i;
> +
> + for ( i = 0; i< *indent; i++ )
> + printk(" ");
> +
> + printk("gfn: %"PRIpaddr" mfn: %"PRIpaddr"\n",
> + PFN_DOWN(address), PFN_DOWN(next_table_maddr));
> +
> + if ( (next_table_maddr != 0)&& (next_level != 0) )
> + {
> + *indent += 1;
> + amd_dump_p2m_table_level(
> + maddr_to_page(next_table_maddr), level - 1,
> + address, indent);
> +
> + *indent -= 1;
> + }
> + }
> + }
> +
> + unmap_domain_page(table_vaddr);
> +}
> +
> +static void amd_dump_p2m_table(struct domain *d)
> +{
> + struct hvm_iommu *hd = domain_hvm_iommu(d);
> + int indent;
> +
> + if ( !hd->root_table )
> + return;
> +
> + indent = 0;
> + amd_dump_p2m_table_level(hd->root_table, hd->paging_mode, 0,&indent);
> +}
> +
> const struct iommu_ops amd_iommu_ops = {
> .init = amd_iommu_domain_init,
> .dom0_init = amd_iommu_dom0_init,
> @@ -531,4 +607,5 @@ const struct iommu_ops amd_iommu_ops = {
> .resume = amd_iommu_resume,
> .share_p2m = amd_iommu_share_p2m,
> .crash_shutdown = amd_iommu_suspend,
> + .dump_p2m_table = amd_dump_p2m_table,
> };
I tested the amd part with an amd iommu system and I got output like this:
XEN) domain1 IOMMU p2m table:
(XEN) gfn: 0000000000000000 mfn: 00000000001b1fb5
(XEN) gfn: 0000000000000000 mfn: 000000000021f80b
(XEN) gfn: 0000000000000000 mfn: 000000000023f400
(XEN) gfn: 0000000000000000 mfn: 000000000010a600
(XEN) gfn: 0000000000000000 mfn: 000000000023f200
(XEN) gfn: 0000000000000000 mfn: 000000000010a400
(XEN) gfn: 0000000000000000 mfn: 000000000023f000
(XEN) gfn: 0000000000000000 mfn: 000000000010ae00
(XEN) gfn: 0000000000000000 mfn: 000000000023ee00
(XEN) gfn: 0000000000000001 mfn: 000000000010ac00
(XEN) gfn: 0000000000000001 mfn: 000000000023ec00
(XEN) gfn: 0000000000000001 mfn: 000000000010aa00
(XEN) gfn: 0000000000000001 mfn: 000000000023ea00
(XEN) gfn: 0000000000000001 mfn: 000000000010a800
(XEN) gfn: 0000000000000001 mfn: 000000000023e800
(XEN) gfn: 0000000000000001 mfn: 000000000010be00
(XEN) gfn: 0000000000000001 mfn: 000000000023e600
(XEN) gfn: 0000000000000002 mfn: 000000000010bc00
(XEN) gfn: 0000000000000002 mfn: 000000000023e400
(XEN) gfn: 0000000000000002 mfn: 000000000010ba00
(XEN) gfn: 0000000000000002 mfn: 000000000023e200
(XEN) gfn: 0000000000000002 mfn: 000000000010b800
(XEN) gfn: 0000000000000002 mfn: 000000000023e000
(XEN) gfn: 0000000000000002 mfn: 000000000010b600
(XEN) gfn: 0000000000000002 mfn: 000000000023de00
(XEN) gfn: 0000000000000003 mfn: 000000000010b400
(XEN) gfn: 0000000000000003 mfn: 000000000023dc00
(XEN) gfn: 0000000000000003 mfn: 000000000010b200
(XEN) gfn: 0000000000000003 mfn: 000000000023da00
(XEN) gfn: 0000000000000003 mfn: 000000000010b000
(XEN) gfn: 0000000000000003 mfn: 000000000023d800
(XEN) gfn: 0000000000000003 mfn: 000000000010fe00
(XEN) gfn: 0000000000000003 mfn: 000000000023d600
(XEN) gfn: 0000000000000004 mfn: 000000000010fc00
(XEN) gfn: 0000000000000004 mfn: 000000000023d400
(XEN) gfn: 0000000000000004 mfn: 000000000010fa00
(XEN) gfn: 0000000000000004 mfn: 000000000023d200
(XEN) gfn: 0000000000000004 mfn: 000000000010f800
(XEN) gfn: 0000000000000004 mfn: 000000000023d000
(XEN) gfn: 0000000000000004 mfn: 000000000010f600
(XEN) gfn: 0000000000000004 mfn: 000000000023ce00
(XEN) gfn: 0000000000000005 mfn: 000000000010f400
(XEN) gfn: 0000000000000005 mfn: 000000000023cc00
(XEN) gfn: 0000000000000005 mfn: 000000000010f200
(XEN) gfn: 0000000000000005 mfn: 000000000023ca00
(XEN) gfn: 0000000000000005 mfn: 000000000010f000
(XEN) gfn: 0000000000000005 mfn: 000000000023c800
(XEN) gfn: 0000000000000005 mfn: 000000000010ee00
(XEN) gfn: 0000000000000005 mfn: 000000000023c600
(XEN) gfn: 0000000000000006 mfn: 000000000010ec00
(XEN) gfn: 0000000000000006 mfn: 000000000023c400
(XEN) gfn: 0000000000000006 mfn: 000000000010ea00
(XEN) gfn: 0000000000000006 mfn: 000000000023c200
(XEN) gfn: 0000000000000006 mfn: 000000000010e800
(XEN) gfn: 0000000000000006 mfn: 000000000023c000
It looks that the same gfn has been mapped to multiple mfns. Do you want
to output only the gfn to mfn mapping or you also want to output the
address of intermediate page tables? What do "gfn" and "mfn" stands for
here?
Thanks,
Wei
> diff -r 472fc515a463 -r f687c5580262 xen/drivers/passthrough/iommu.c
> --- a/xen/drivers/passthrough/iommu.c Tue Aug 07 18:37:31 2012 +0100
> +++ b/xen/drivers/passthrough/iommu.c Thu Aug 09 18:31:32 2012 -0700
> @@ -18,11 +18,13 @@
> #include<asm/hvm/iommu.h>
> #include<xen/paging.h>
> #include<xen/guest_access.h>
> +#include<xen/keyhandler.h>
> #include<xen/softirq.h>
> #include<xsm/xsm.h>
>
> static void parse_iommu_param(char *s);
> static int iommu_populate_page_table(struct domain *d);
> +static void iommu_dump_p2m_table(unsigned char key);
>
> /*
> * The 'iommu' parameter enables the IOMMU. Optional comma separated
> @@ -54,6 +56,12 @@ bool_t __read_mostly amd_iommu_perdev_in
>
> DEFINE_PER_CPU(bool_t, iommu_dont_flush_iotlb);
>
> +static struct keyhandler iommu_p2m_table = {
> + .diagnostic = 0,
> + .u.fn = iommu_dump_p2m_table,
> + .desc = "dump iommu p2m table"
> +};
> +
> static void __init parse_iommu_param(char *s)
> {
> char *ss;
> @@ -119,6 +127,7 @@ void __init iommu_dom0_init(struct domai
> if ( !iommu_enabled )
> return;
>
> + register_keyhandler('o',&iommu_p2m_table);
> d->need_iommu = !!iommu_dom0_strict;
> if ( need_iommu(d) )
> {
> @@ -654,6 +663,34 @@ int iommu_do_domctl(
> return ret;
> }
>
> +static void iommu_dump_p2m_table(unsigned char key)
> +{
> + struct domain *d;
> + const struct iommu_ops *ops;
> +
> + if ( !iommu_enabled )
> + {
> + printk("IOMMU not enabled!\n");
> + return;
> + }
> +
> + ops = iommu_get_ops();
> + for_each_domain(d)
> + {
> + if ( !d->domain_id )
> + continue;
> +
> + if ( iommu_use_hap_pt(d) )
> + {
> + printk("\ndomain%d IOMMU p2m table shared with MMU: \n", d->domain_id);
> + continue;
> + }
> +
> + printk("\ndomain%d IOMMU p2m table: \n", d->domain_id);
> + ops->dump_p2m_table(d);
> + }
> +}
> +
> /*
> * Local variables:
> * mode: C
> diff -r 472fc515a463 -r f687c5580262 xen/drivers/passthrough/vtd/iommu.c
> --- a/xen/drivers/passthrough/vtd/iommu.c Tue Aug 07 18:37:31 2012 +0100
> +++ b/xen/drivers/passthrough/vtd/iommu.c Thu Aug 09 18:31:32 2012 -0700
> @@ -31,6 +31,7 @@
> #include<xen/pci.h>
> #include<xen/pci_regs.h>
> #include<xen/keyhandler.h>
> +#include<xen/softirq.h>
> #include<asm/msi.h>
> #include<asm/irq.h>
> #if defined(__i386__) || defined(__x86_64__)
> @@ -2365,6 +2366,71 @@ static void vtd_resume(void)
> }
> }
>
> +static void vtd_dump_p2m_table_level(paddr_t pt_maddr, int level, paddr_t gpa,
> + int *indent)
> +{
> + paddr_t address;
> + int i;
> + struct dma_pte *pt_vaddr, *pte;
> + int next_level;
> +
> + if ( pt_maddr == 0 )
> + return;
> +
> + pt_vaddr = map_vtd_domain_page(pt_maddr);
> + if ( pt_vaddr == NULL )
> + {
> + printk("Failed to map VT-D domain page %"PRIpaddr"\n", pt_maddr);
> + return;
> + }
> +
> + next_level = level - 1;
> + for ( i = 0; i< PTE_NUM; i++ )
> + {
> + if ( !(i % 2) )
> + process_pending_softirqs();
> +
> + pte =&pt_vaddr[i];
> + if ( dma_pte_present(*pte) )
> + {
> + int j;
> +
> + for ( j = 0; j< *indent; j++ )
> + printk(" ");
> +
> + address = gpa + offset_level_address(i, level);
> + printk("gfn: %"PRIpaddr" mfn: %"PRIpaddr" super=%d rd=%d wr=%d\n",
> + address>> PAGE_SHIFT_4K, pte->val>> PAGE_SHIFT_4K,
> + dma_pte_superpage(*pte)? 1 : 0, dma_pte_read(*pte)? 1 : 0,
> + dma_pte_write(*pte)? 1 : 0);
> +
> + if ( next_level>= 1 ) {
> + *indent += 1;
> + vtd_dump_p2m_table_level(dma_pte_addr(*pte), next_level,
> + address, indent);
> +
> + *indent -= 1;
> + }
> + }
> + }
> +
> + unmap_vtd_domain_page(pt_vaddr);
> +}
> +
> +static void vtd_dump_p2m_table(struct domain *d)
> +{
> + struct hvm_iommu *hd;
> + int indent;
> +
> + if ( list_empty(&acpi_drhd_units) )
> + return;
> +
> + hd = domain_hvm_iommu(d);
> + indent = 0;
> + vtd_dump_p2m_table_level(hd->pgd_maddr, agaw_to_level(hd->agaw),
> + 0,&indent);
> +}
> +
> const struct iommu_ops intel_iommu_ops = {
> .init = intel_iommu_domain_init,
> .dom0_init = intel_iommu_dom0_init,
> @@ -2387,6 +2453,7 @@ const struct iommu_ops intel_iommu_ops =
> .crash_shutdown = vtd_crash_shutdown,
> .iotlb_flush = intel_iommu_iotlb_flush,
> .iotlb_flush_all = intel_iommu_iotlb_flush_all,
> + .dump_p2m_table = vtd_dump_p2m_table,
> };
>
> /*
> diff -r 472fc515a463 -r f687c5580262 xen/drivers/passthrough/vtd/iommu.h
> --- a/xen/drivers/passthrough/vtd/iommu.h Tue Aug 07 18:37:31 2012 +0100
> +++ b/xen/drivers/passthrough/vtd/iommu.h Thu Aug 09 18:31:32 2012 -0700
> @@ -248,6 +248,8 @@ struct context_entry {
> #define level_to_offset_bits(l) (12 + (l - 1) * LEVEL_STRIDE)
> #define address_level_offset(addr, level) \
> ((addr>> level_to_offset_bits(level))& LEVEL_MASK)
> +#define offset_level_address(offset, level) \
> + ((u64)(offset)<< level_to_offset_bits(level))
> #define level_mask(l) (((u64)(-1))<< level_to_offset_bits(l))
> #define level_size(l) (1<< level_to_offset_bits(l))
> #define align_to_level(addr, l) ((addr + level_size(l) - 1)& level_mask(l))
> @@ -277,6 +279,9 @@ struct dma_pte {
> #define dma_set_pte_addr(p, addr) do {\
> (p).val |= ((addr)& PAGE_MASK_4K); } while (0)
> #define dma_pte_present(p) (((p).val& 3) != 0)
> +#define dma_pte_superpage(p) (((p).val& (1<<7)) != 0)
> +#define dma_pte_read(p) (((p).val& DMA_PTE_READ) != 0)
> +#define dma_pte_write(p) (((p).val& DMA_PTE_WRITE) != 0)
>
> /* interrupt remap entry */
> struct iremap_entry {
> diff -r 472fc515a463 -r f687c5580262 xen/include/asm-x86/hvm/svm/amd-iommu-defs.h
> --- a/xen/include/asm-x86/hvm/svm/amd-iommu-defs.h Tue Aug 07 18:37:31 2012 +0100
> +++ b/xen/include/asm-x86/hvm/svm/amd-iommu-defs.h Thu Aug 09 18:31:32 2012 -0700
> @@ -38,6 +38,10 @@
> #define PTE_PER_TABLE_ALLOC(entries) \
> PAGE_SIZE * (PTE_PER_TABLE_ALIGN(entries)>> PTE_PER_TABLE_SHIFT)
>
> +#define amd_offset_level_address(offset, level) \
> + ((u64)(offset)<< ((PTE_PER_TABLE_SHIFT * \
> + (level - IOMMU_PAGING_MODE_LEVEL_1))))
> +
> #define PCI_MIN_CAP_OFFSET 0x40
> #define PCI_MAX_CAP_BLOCKS 48
> #define PCI_CAP_PTR_MASK 0xFC
> diff -r 472fc515a463 -r f687c5580262 xen/include/xen/iommu.h
> --- a/xen/include/xen/iommu.h Tue Aug 07 18:37:31 2012 +0100
> +++ b/xen/include/xen/iommu.h Thu Aug 09 18:31:32 2012 -0700
> @@ -141,6 +141,7 @@ struct iommu_ops {
> void (*crash_shutdown)(void);
> void (*iotlb_flush)(struct domain *d, unsigned long gfn, unsigned int page_count);
> void (*iotlb_flush_all)(struct domain *d);
> + void (*dump_p2m_table)(struct domain *d);
> };
>
> void iommu_update_ire_from_apic(unsigned int apic, unsigned int reg, unsigned int value);
>
next prev parent reply other threads:[~2012-08-10 12:31 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-08-10 1:43 [PATCH] dump_p2m_table: For IOMMU Santosh Jodh
2012-08-10 7:49 ` Jan Beulich
2012-08-10 12:31 ` Wei Wang [this message]
2012-08-10 13:02 ` Jan Beulich
2012-08-10 15:02 ` Santosh Jodh
-- strict thread matches above, loose matches on Subject: below --
2012-08-10 19:14 Santosh Jodh
2012-08-13 8:59 ` Jan Beulich
2012-08-14 19:34 ` Santosh Jodh
2012-08-13 10:31 ` Wei Wang
2012-08-08 17:17 Santosh Jodh
2012-08-09 7:26 ` Jan Beulich
2012-08-10 1:41 ` Santosh Jodh
2012-08-10 10:50 ` Wei Wang
2012-08-10 12:52 ` Jan Beulich
2012-08-10 13:41 ` Wei Wang
2012-08-10 14:24 ` Jan Beulich
2012-08-08 15:56 Santosh Jodh
2012-08-08 16:21 ` Jan Beulich
2012-08-08 17:17 ` Santosh Jodh
2012-08-07 14:49 Santosh Jodh
2012-08-07 15:52 ` Jan Beulich
2012-08-07 16:16 ` Santosh Jodh
2012-08-08 7:31 ` Jan Beulich
2012-08-08 15:32 ` Santosh Jodh
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=5024FF0A.6090006@amd.com \
--to=wei.wang2@amd.com \
--cc=santosh.jodh@citrix.com \
--cc=tim@xen.org \
--cc=xen-devel@lists.xensource.com \
--cc=xiantao.zhang@intel.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;
as well as URLs for NNTP newsgroup(s).