Re: [PATCH 10/10] xen/swiotlb: Depending on after_bootmem is notcorrect.

From: Konrad Rzeszutek Wilk
Date: Fri Sep 14 2012 - 13:20:13 EST


On Fri, Sep 14, 2012 at 05:10:48PM +0100, Stefano Stabellini wrote:
> On Mon, 10 Sep 2012, Konrad Rzeszutek Wilk wrote:
> > When PCI IOMMUs are initialized it is after after_bootmem but
> > before a lot of "other" subsystems are initialized. As such
> > the check for after_bootmem is incorrect and we should
> > just use a parameter to define whether we are early or late.
> >
> > This solves this bootup problem:
> >
> > __ex_table already sorted, skipping sortM
> > Initializing CPU#0
> > Warning: only able to allocate 1 MB for software IO TLB
> > Xen-SWIOTLB: Lowering to 2MB
> > Warning: only able to allocate 1 MB for software IO TLB
> > Xen-SWIOTLB: Lowering to 2MB
> > Warning: only able to allocate 1 MB for software IO TLB
> > Xen-SWIOTLB: Lowering to 2MB
> > Warning: only able to allocate 1 MB for software IO TLB
> > Cannot allocate Xen-SWIOTLB buffer (rc:-12)
> >
> > Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
> > ---
> > arch/x86/xen/pci-swiotlb-xen.c | 4 ++--
> > drivers/xen/swiotlb-xen.c | 11 ++++++-----
> > include/xen/swiotlb-xen.h | 2 +-
> > 3 files changed, 9 insertions(+), 8 deletions(-)
> >
> > diff --git a/arch/x86/xen/pci-swiotlb-xen.c b/arch/x86/xen/pci-swiotlb-xen.c
> > index fc0c78f..1608244 100644
> > --- a/arch/x86/xen/pci-swiotlb-xen.c
> > +++ b/arch/x86/xen/pci-swiotlb-xen.c
> > @@ -69,7 +69,7 @@ int __init pci_xen_swiotlb_detect(void)
> > void __init pci_xen_swiotlb_init(void)
> > {
> > if (xen_swiotlb) {
> > - xen_swiotlb_init(1);
> > + xen_swiotlb_init(1, true /* early */);
> > dma_ops = &xen_swiotlb_dma_ops;
> >
> > /* Make sure ACS will be enabled */
> > @@ -84,7 +84,7 @@ int pci_xen_swiotlb_init_late(void)
> > if (xen_swiotlb)
> > return 0;
> >
> > - rc = xen_swiotlb_init(1);
> > + rc = xen_swiotlb_init(1, false /* late */);
> > if (rc)
> > return rc;
> >
> > diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
> > index 6f81994..7443e19 100644
> > --- a/drivers/xen/swiotlb-xen.c
> > +++ b/drivers/xen/swiotlb-xen.c
> > @@ -176,7 +176,7 @@ static const char *xen_swiotlb_error(enum xen_swiotlb_err err)
> > }
> > return "";
> > }
> > -int __ref xen_swiotlb_init(int verbose)
> > +int __ref xen_swiotlb_init(int verbose, bool early)
> > {
> > unsigned long bytes, order;
> > int rc = -ENOMEM;
> > @@ -190,7 +190,7 @@ retry:
> > /*
> > * Get IO TLB memory from any location.
> > */
> > - if (!after_bootmem)
> > + if (early)
> > xen_io_tlb_start = alloc_bootmem_pages(PAGE_ALIGN(bytes));
> > else {
> > #define SLABS_PER_PAGE (1 << (PAGE_SHIFT - IO_TLB_SHIFT))
> > @@ -220,7 +220,7 @@ retry:
> > bytes,
> > xen_io_tlb_nslabs);
> > if (rc) {
> > - if (!after_bootmem)
> > + if (early)
> > free_bootmem(__pa(xen_io_tlb_start), PAGE_ALIGN(bytes));
> > else {
> > free_pages((unsigned long)xen_io_tlb_start, order);
> > @@ -230,7 +230,8 @@ retry:
> > goto error;
> > }
> > start_dma_addr = xen_virt_to_bus(xen_io_tlb_start);
> > - if (!after_bootmem)
> > + rc = 0;
> ^
> why does this change belong to this patch?

Good eyes. I was being sneaky and rolled it in. But I can't recall why I needed
this. <sigh>

>
>
> > + if (early)
> > swiotlb_init_with_tbl(xen_io_tlb_start, xen_io_tlb_nslabs, verbose);
> > else
> > rc = swiotlb_late_init_with_tbl(xen_io_tlb_start, xen_io_tlb_nslabs);
> > @@ -244,7 +245,7 @@ error:
> > goto retry;
> > }
> > pr_err("%s (rc:%d)", xen_swiotlb_error(m_ret), rc);
> > - if (!after_bootmem)
> > + if (early)
> > panic("%s (rc:%d)", xen_swiotlb_error(m_ret), rc);
> > else
> > free_pages((unsigned long)xen_io_tlb_start, order);
> > diff --git a/include/xen/swiotlb-xen.h b/include/xen/swiotlb-xen.h
> > index a0db2b7..de8bcc6 100644
> > --- a/include/xen/swiotlb-xen.h
> > +++ b/include/xen/swiotlb-xen.h
> > @@ -3,7 +3,7 @@
> >
> > #include <linux/swiotlb.h>
> >
> > -extern int xen_swiotlb_init(int verbose);
> > +extern int xen_swiotlb_init(int verbose, bool early);
> >
> > extern void
> > *xen_swiotlb_alloc_coherent(struct device *hwdev, size_t size,
> > --
> > 1.7.7.6
> >
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/