* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
[not found] ` <20170930161023.GI2965@mtr-leonro.local>
@ 2017-10-12 12:31 ` Wei Hu (Xavier)
2017-10-12 12:59 ` Robin Murphy
2017-10-12 14:54 ` Leon Romanovsky
0 siblings, 2 replies; 14+ messages in thread
From: Wei Hu (Xavier) @ 2017-10-12 12:31 UTC (permalink / raw)
To: Leon Romanovsky
Cc: dledford, linux-rdma, lijun_nudt, oulijun, charles.chenxin,
liuyixian, linux-mm, zhangxiping3, xavier.huwei, linuxarm,
linux-kernel, shaobo.xu, shaoboxu, leizhen 00275356, joro, iommu
On 2017/10/1 0:10, Leon Romanovsky wrote:
> On Sat, Sep 30, 2017 at 05:28:59PM +0800, Wei Hu (Xavier) wrote:
>> If the IOMMU is enabled, the length of sg obtained from
>> __iommu_map_sg_attrs is not 4kB. When the IOVA is set with the sg
>> dma address, the IOVA will not be page continuous. and the VA
>> returned from dma_alloc_coherent is a vmalloc address. However,
>> the VA obtained by the page_address is a discontinuous VA. Under
>> these circumstances, the IOVA should be calculated based on the
>> sg length, and record the VA returned from dma_alloc_coherent
>> in the struct of hem.
>>
>> Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
>> Signed-off-by: Shaobo Xu <xushaobo2@huawei.com>
>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>> ---
> Doug,
>
> I didn't invest time in reviewing it, but having "is_vmalloc_addr" in
> driver code to deal with dma_alloc_coherent is most probably wrong.
>
> Thanks
Hi, Leon & Doug
We refered the function named __ttm_dma_alloc_page in the kernel
code as below:
And there are similar methods in bch_bio_map and mem_to_page
functions in current 4.14-rcx.
static struct dma_page *__ttm_dma_alloc_page(struct dma_pool *pool)
{
struct dma_page *d_page;
d_page = kmalloc(sizeof(struct dma_page), GFP_KERNEL);
if (!d_page)
return NULL;
d_page->vaddr = dma_alloc_coherent(pool->dev, pool->size,
&d_page->dma,
pool->gfp_flags);
if (d_page->vaddr) {
if (is_vmalloc_addr(d_page->vaddr))
d_page->p = vmalloc_to_page(d_page->vaddr);
else
d_page->p = virt_to_page(d_page->vaddr);
} else {
kfree(d_page);
d_page = NULL;
}
return d_page;
}
Regards
Wei Hu
>
>> drivers/infiniband/hw/hns/hns_roce_alloc.c | 5 ++++-
>> drivers/infiniband/hw/hns/hns_roce_hem.c | 30 +++++++++++++++++++++++++++---
>> drivers/infiniband/hw/hns/hns_roce_hem.h | 6 ++++++
>> drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 22 +++++++++++++++-------
>> 4 files changed, 52 insertions(+), 11 deletions(-)
>>
>> diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>> index 3e4c525..a69cd4b 100644
>> --- a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>> +++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>> @@ -243,7 +243,10 @@ int hns_roce_buf_alloc(struct hns_roce_dev *hr_dev, u32 size, u32 max_direct,
>> goto err_free;
>>
>> for (i = 0; i < buf->nbufs; ++i)
>> - pages[i] = virt_to_page(buf->page_list[i].buf);
>> + pages[i] =
>> + is_vmalloc_addr(buf->page_list[i].buf) ?
>> + vmalloc_to_page(buf->page_list[i].buf) :
>> + virt_to_page(buf->page_list[i].buf);
>>
>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>> PAGE_KERNEL);
>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c b/drivers/infiniband/hw/hns/hns_roce_hem.c
>> index 8388ae2..4a3d1d4 100644
>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.c
>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c
>> @@ -200,6 +200,7 @@ static struct hns_roce_hem *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>> gfp_t gfp_mask)
>> {
>> struct hns_roce_hem_chunk *chunk = NULL;
>> + struct hns_roce_vmalloc *vmalloc;
>> struct hns_roce_hem *hem;
>> struct scatterlist *mem;
>> int order;
>> @@ -227,6 +228,7 @@ static struct hns_roce_hem *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>> sg_init_table(chunk->mem, HNS_ROCE_HEM_CHUNK_LEN);
>> chunk->npages = 0;
>> chunk->nsg = 0;
>> + memset(chunk->vmalloc, 0, sizeof(chunk->vmalloc));
>> list_add_tail(&chunk->list, &hem->chunk_list);
>> }
>>
>> @@ -243,7 +245,15 @@ static struct hns_roce_hem *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>> if (!buf)
>> goto fail;
>>
>> - sg_set_buf(mem, buf, PAGE_SIZE << order);
>> + if (is_vmalloc_addr(buf)) {
>> + vmalloc = &chunk->vmalloc[chunk->npages];
>> + vmalloc->is_vmalloc_addr = true;
>> + vmalloc->vmalloc_addr = buf;
>> + sg_set_page(mem, vmalloc_to_page(buf),
>> + PAGE_SIZE << order, offset_in_page(buf));
>> + } else {
>> + sg_set_buf(mem, buf, PAGE_SIZE << order);
>> + }
>> WARN_ON(mem->offset);
>> sg_dma_len(mem) = PAGE_SIZE << order;
>>
>> @@ -262,17 +272,25 @@ static struct hns_roce_hem *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>> void hns_roce_free_hem(struct hns_roce_dev *hr_dev, struct hns_roce_hem *hem)
>> {
>> struct hns_roce_hem_chunk *chunk, *tmp;
>> + void *cpu_addr;
>> int i;
>>
>> if (!hem)
>> return;
>>
>> list_for_each_entry_safe(chunk, tmp, &hem->chunk_list, list) {
>> - for (i = 0; i < chunk->npages; ++i)
>> + for (i = 0; i < chunk->npages; ++i) {
>> + if (chunk->vmalloc[i].is_vmalloc_addr)
>> + cpu_addr = chunk->vmalloc[i].vmalloc_addr;
>> + else
>> + cpu_addr =
>> + lowmem_page_address(sg_page(&chunk->mem[i]));
>> +
>> dma_free_coherent(hr_dev->dev,
>> chunk->mem[i].length,
>> - lowmem_page_address(sg_page(&chunk->mem[i])),
>> + cpu_addr,
>> sg_dma_address(&chunk->mem[i]));
>> + }
>> kfree(chunk);
>> }
>>
>> @@ -774,6 +792,12 @@ void *hns_roce_table_find(struct hns_roce_dev *hr_dev,
>>
>> if (chunk->mem[i].length > (u32)offset) {
>> page = sg_page(&chunk->mem[i]);
>> + if (chunk->vmalloc[i].is_vmalloc_addr) {
>> + mutex_unlock(&table->mutex);
>> + return page ?
>> + chunk->vmalloc[i].vmalloc_addr
>> + + offset : NULL;
>> + }
>> goto out;
>> }
>> offset -= chunk->mem[i].length;
>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h b/drivers/infiniband/hw/hns/hns_roce_hem.h
>> index af28bbf..62d712a 100644
>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.h
>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h
>> @@ -72,11 +72,17 @@ enum {
>> HNS_ROCE_HEM_PAGE_SIZE = 1 << HNS_ROCE_HEM_PAGE_SHIFT,
>> };
>>
>> +struct hns_roce_vmalloc {
>> + bool is_vmalloc_addr;
>> + void *vmalloc_addr;
>> +};
>> +
>> struct hns_roce_hem_chunk {
>> struct list_head list;
>> int npages;
>> int nsg;
>> struct scatterlist mem[HNS_ROCE_HEM_CHUNK_LEN];
>> + struct hns_roce_vmalloc vmalloc[HNS_ROCE_HEM_CHUNK_LEN];
>> };
>>
>> struct hns_roce_hem {
>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>> index b99d70a..9e19bf1 100644
>> --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>> +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>> @@ -1093,9 +1093,11 @@ static int hns_roce_v2_write_mtpt(void *mb_buf, struct hns_roce_mr *mr,
>> {
>> struct hns_roce_v2_mpt_entry *mpt_entry;
>> struct scatterlist *sg;
>> + u64 page_addr = 0;
>> u64 *pages;
>> + int i = 0, j = 0;
>> + int len = 0;
>> int entry;
>> - int i;
>>
>> mpt_entry = mb_buf;
>> memset(mpt_entry, 0, sizeof(*mpt_entry));
>> @@ -1153,14 +1155,20 @@ static int hns_roce_v2_write_mtpt(void *mb_buf, struct hns_roce_mr *mr,
>>
>> i = 0;
>> for_each_sg(mr->umem->sg_head.sgl, sg, mr->umem->nmap, entry) {
>> - pages[i] = ((u64)sg_dma_address(sg)) >> 6;
>> -
>> - /* Record the first 2 entry directly to MTPT table */
>> - if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>> - break;
>> - i++;
>> + len = sg_dma_len(sg) >> PAGE_SHIFT;
>> + for (j = 0; j < len; ++j) {
>> + page_addr = sg_dma_address(sg) +
>> + (j << mr->umem->page_shift);
>> + pages[i] = page_addr >> 6;
>> +
>> + /* Record the first 2 entry directly to MTPT table */
>> + if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>> + goto found;
>> + i++;
>> + }
>> }
>>
>> +found:
>> mpt_entry->pa0_l = cpu_to_le32(lower_32_bits(pages[0]));
>> roce_set_field(mpt_entry->byte_56_pa0_h, V2_MPT_BYTE_56_PA0_H_M,
>> V2_MPT_BYTE_56_PA0_H_S,
>> --
>> 1.9.1
>>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-10-12 12:31 ` [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08 Wei Hu (Xavier)
@ 2017-10-12 12:59 ` Robin Murphy
2017-11-01 7:46 ` Wei Hu (Xavier)
2017-10-12 14:54 ` Leon Romanovsky
1 sibling, 1 reply; 14+ messages in thread
From: Robin Murphy @ 2017-10-12 12:59 UTC (permalink / raw)
To: Wei Hu (Xavier), Leon Romanovsky
Cc: shaobo.xu, xavier.huwei, lijun_nudt, oulijun, linux-rdma,
charles.chenxin, linuxarm, iommu, linux-kernel, linux-mm,
dledford, liuyixian, zhangxiping3, shaoboxu
On 12/10/17 13:31, Wei Hu (Xavier) wrote:
>
>
> On 2017/10/1 0:10, Leon Romanovsky wrote:
>> On Sat, Sep 30, 2017 at 05:28:59PM +0800, Wei Hu (Xavier) wrote:
>>> If the IOMMU is enabled, the length of sg obtained from
>>> __iommu_map_sg_attrs is not 4kB. When the IOVA is set with the sg
>>> dma address, the IOVA will not be page continuous. and the VA
>>> returned from dma_alloc_coherent is a vmalloc address. However,
>>> the VA obtained by the page_address is a discontinuous VA. Under
>>> these circumstances, the IOVA should be calculated based on the
>>> sg length, and record the VA returned from dma_alloc_coherent
>>> in the struct of hem.
>>>
>>> Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
>>> Signed-off-by: Shaobo Xu <xushaobo2@huawei.com>
>>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>>> ---
>> Doug,
>>
>> I didn't invest time in reviewing it, but having "is_vmalloc_addr" in
>> driver code to deal with dma_alloc_coherent is most probably wrong.
>>
>> Thanks
> Hi,A Leon & Doug
> A A A We refered the function named __ttm_dma_alloc_page in the kernel
> code as below:
> A A A And there are similar methods in bch_bio_map and mem_to_page
> functions in current 4.14-rcx.
>
> A A A A A A A static struct dma_page *__ttm_dma_alloc_page(struct dma_pool *pool)
> A A A A A A A {
> A A A A A A A A A A A struct dma_page *d_page;
>
> A A A A A A A A A A A d_page = kmalloc(sizeof(struct dma_page), GFP_KERNEL);
> A A A A A A A A A A A if (!d_page)
> A A A A A A A A A A A A A A A return NULL;
>
> A A A A A A A A A A A d_page->vaddr = dma_alloc_coherent(pool->dev, pool->size,
> A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A &d_page->dma,
> A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A A pool->gfp_flags);
> A A A A A A A A A A A if (d_page->vaddr) {
> A A A A A A A A A A A A A A A if (is_vmalloc_addr(d_page->vaddr))
> A A A A A A A A A A A A A A A A A A A d_page->p = vmalloc_to_page(d_page->vaddr);
> A A A A A A A A A A A A A A A else
> A A A A A A A A A A A A A A A A A A A d_page->p = virt_to_page(d_page->vaddr);
There are cases on various architectures where neither of those is
right. Whether those actually intersect with TTM or RDMA use-cases is
another matter, of course.
What definitely is a problem is if you ever take that page and end up
accessing it through any virtual address other than the one explicitly
returned by dma_alloc_coherent(). That can blow the coherency wide open
and invite data loss, right up to killing the whole system with a
machine check on certain architectures.
Robin.
> A A A A A A A A A A A } else {
> A A A A A A A A A A A A A A A kfree(d_page);
> A A A A A A A A A A A A A A A d_page = NULL;
> A A A A A A A A A A A }
> A A A A A A A A A A A return d_page;
> A A A A A A A }
>
> A A A Regards
> Wei Hu
>>
>>> A drivers/infiniband/hw/hns/hns_roce_alloc.c |A 5 ++++-
>>> A drivers/infiniband/hw/hns/hns_roce_hem.cA A | 30
>>> +++++++++++++++++++++++++++---
>>> A drivers/infiniband/hw/hns/hns_roce_hem.hA A |A 6 ++++++
>>> A drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 22 +++++++++++++++-------
>>> A 4 files changed, 52 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>> b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>> index 3e4c525..a69cd4b 100644
>>> --- a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>> +++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>> @@ -243,7 +243,10 @@ int hns_roce_buf_alloc(struct hns_roce_dev
>>> *hr_dev, u32 size, u32 max_direct,
>>> A A A A A A A A A A A A A A A A A goto err_free;
>>>
>>> A A A A A A A A A A A A A for (i = 0; i < buf->nbufs; ++i)
>>> -A A A A A A A A A A A A A A A pages[i] = virt_to_page(buf->page_list[i].buf);
>>> +A A A A A A A A A A A A A A A pages[i] =
>>> +A A A A A A A A A A A A A A A A A A A is_vmalloc_addr(buf->page_list[i].buf) ?
>>> +A A A A A A A A A A A A A A A A A A A vmalloc_to_page(buf->page_list[i].buf) :
>>> +A A A A A A A A A A A A A A A A A A A virt_to_page(buf->page_list[i].buf);
>>>
>>> A A A A A A A A A A A A A buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>>> A A A A A A A A A A A A A A A A A A A A A A A A A A A A PAGE_KERNEL);
>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>> b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>> index 8388ae2..4a3d1d4 100644
>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>> @@ -200,6 +200,7 @@ static struct hns_roce_hem
>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>> A A A A A A A A A A A A A A A A A A A A A A A A A A A A gfp_t gfp_mask)
>>> A {
>>> A A A A A struct hns_roce_hem_chunk *chunk = NULL;
>>> +A A A struct hns_roce_vmalloc *vmalloc;
>>> A A A A A struct hns_roce_hem *hem;
>>> A A A A A struct scatterlist *mem;
>>> A A A A A int order;
>>> @@ -227,6 +228,7 @@ static struct hns_roce_hem
>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>> A A A A A A A A A A A A A sg_init_table(chunk->mem, HNS_ROCE_HEM_CHUNK_LEN);
>>> A A A A A A A A A A A A A chunk->npages = 0;
>>> A A A A A A A A A A A A A chunk->nsg = 0;
>>> +A A A A A A A A A A A memset(chunk->vmalloc, 0, sizeof(chunk->vmalloc));
>>> A A A A A A A A A A A A A list_add_tail(&chunk->list, &hem->chunk_list);
>>> A A A A A A A A A }
>>>
>>> @@ -243,7 +245,15 @@ static struct hns_roce_hem
>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>> A A A A A A A A A if (!buf)
>>> A A A A A A A A A A A A A goto fail;
>>>
>>> -A A A A A A A sg_set_buf(mem, buf, PAGE_SIZE << order);
>>> +A A A A A A A if (is_vmalloc_addr(buf)) {
>>> +A A A A A A A A A A A vmalloc = &chunk->vmalloc[chunk->npages];
>>> +A A A A A A A A A A A vmalloc->is_vmalloc_addr = true;
>>> +A A A A A A A A A A A vmalloc->vmalloc_addr = buf;
>>> +A A A A A A A A A A A sg_set_page(mem, vmalloc_to_page(buf),
>>> +A A A A A A A A A A A A A A A A A A A PAGE_SIZE << order, offset_in_page(buf));
>>> +A A A A A A A } else {
>>> +A A A A A A A A A A A sg_set_buf(mem, buf, PAGE_SIZE << order);
>>> +A A A A A A A }
>>> A A A A A A A A A WARN_ON(mem->offset);
>>> A A A A A A A A A sg_dma_len(mem) = PAGE_SIZE << order;
>>>
>>> @@ -262,17 +272,25 @@ static struct hns_roce_hem
>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>> A void hns_roce_free_hem(struct hns_roce_dev *hr_dev, struct
>>> hns_roce_hem *hem)
>>> A {
>>> A A A A A struct hns_roce_hem_chunk *chunk, *tmp;
>>> +A A A void *cpu_addr;
>>> A A A A A int i;
>>>
>>> A A A A A if (!hem)
>>> A A A A A A A A A return;
>>>
>>> A A A A A list_for_each_entry_safe(chunk, tmp, &hem->chunk_list, list) {
>>> -A A A A A A A for (i = 0; i < chunk->npages; ++i)
>>> +A A A A A A A for (i = 0; i < chunk->npages; ++i) {
>>> +A A A A A A A A A A A if (chunk->vmalloc[i].is_vmalloc_addr)
>>> +A A A A A A A A A A A A A A A cpu_addr = chunk->vmalloc[i].vmalloc_addr;
>>> +A A A A A A A A A A A else
>>> +A A A A A A A A A A A A A A A cpu_addr =
>>> +A A A A A A A A A A A A A A A A A A lowmem_page_address(sg_page(&chunk->mem[i]));
>>> +
>>> A A A A A A A A A A A A A dma_free_coherent(hr_dev->dev,
>>> A A A A A A A A A A A A A A A A A A A A chunk->mem[i].length,
>>> -A A A A A A A A A A A A A A A A A A lowmem_page_address(sg_page(&chunk->mem[i])),
>>> +A A A A A A A A A A A A A A A A A A cpu_addr,
>>> A A A A A A A A A A A A A A A A A A A A sg_dma_address(&chunk->mem[i]));
>>> +A A A A A A A }
>>> A A A A A A A A A kfree(chunk);
>>> A A A A A }
>>>
>>> @@ -774,6 +792,12 @@ void *hns_roce_table_find(struct hns_roce_dev
>>> *hr_dev,
>>>
>>> A A A A A A A A A A A A A if (chunk->mem[i].length > (u32)offset) {
>>> A A A A A A A A A A A A A A A A A page = sg_page(&chunk->mem[i]);
>>> +A A A A A A A A A A A A A A A if (chunk->vmalloc[i].is_vmalloc_addr) {
>>> +A A A A A A A A A A A A A A A A A A A mutex_unlock(&table->mutex);
>>> +A A A A A A A A A A A A A A A A A A A return page ?
>>> +A A A A A A A A A A A A A A A A A A A A A A A chunk->vmalloc[i].vmalloc_addr
>>> +A A A A A A A A A A A A A A A A A A A A A A A + offset : NULL;
>>> +A A A A A A A A A A A A A A A }
>>> A A A A A A A A A A A A A A A A A goto out;
>>> A A A A A A A A A A A A A }
>>> A A A A A A A A A A A A A offset -= chunk->mem[i].length;
>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>> b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>> index af28bbf..62d712a 100644
>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>> @@ -72,11 +72,17 @@ enum {
>>> A A A A A A HNS_ROCE_HEM_PAGE_SIZEA = 1 << HNS_ROCE_HEM_PAGE_SHIFT,
>>> A };
>>>
>>> +struct hns_roce_vmalloc {
>>> +A A A boolA A A is_vmalloc_addr;
>>> +A A A voidA A A *vmalloc_addr;
>>> +};
>>> +
>>> A struct hns_roce_hem_chunk {
>>> A A A A A struct list_headA A A A list;
>>> A A A A A intA A A A A A A A A A A A npages;
>>> A A A A A intA A A A A A A A A A A A nsg;
>>> A A A A A struct scatterlistA A A A mem[HNS_ROCE_HEM_CHUNK_LEN];
>>> +A A A struct hns_roce_vmallocA A A A vmalloc[HNS_ROCE_HEM_CHUNK_LEN];
>>> A };
>>>
>>> A struct hns_roce_hem {
>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>> b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>> index b99d70a..9e19bf1 100644
>>> --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>> @@ -1093,9 +1093,11 @@ static int hns_roce_v2_write_mtpt(void
>>> *mb_buf, struct hns_roce_mr *mr,
>>> A {
>>> A A A A A struct hns_roce_v2_mpt_entry *mpt_entry;
>>> A A A A A struct scatterlist *sg;
>>> +A A A u64 page_addr = 0;
>>> A A A A A u64 *pages;
>>> +A A A int i = 0, j = 0;
>>> +A A A int len = 0;
>>> A A A A A int entry;
>>> -A A A int i;
>>>
>>> A A A A A mpt_entry = mb_buf;
>>> A A A A A memset(mpt_entry, 0, sizeof(*mpt_entry));
>>> @@ -1153,14 +1155,20 @@ static int hns_roce_v2_write_mtpt(void
>>> *mb_buf, struct hns_roce_mr *mr,
>>>
>>> A A A A A i = 0;
>>> A A A A A for_each_sg(mr->umem->sg_head.sgl, sg, mr->umem->nmap, entry) {
>>> -A A A A A A A pages[i] = ((u64)sg_dma_address(sg)) >> 6;
>>> -
>>> -A A A A A A A /* Record the first 2 entry directly to MTPT table */
>>> -A A A A A A A if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>> -A A A A A A A A A A A break;
>>> -A A A A A A A i++;
>>> +A A A A A A A len = sg_dma_len(sg) >> PAGE_SHIFT;
>>> +A A A A A A A for (j = 0; j < len; ++j) {
>>> +A A A A A A A A A A A page_addr = sg_dma_address(sg) +
>>> +A A A A A A A A A A A A A A A A A A A (j << mr->umem->page_shift);
>>> +A A A A A A A A A A A pages[i] = page_addr >> 6;
>>> +
>>> +A A A A A A A A A A A /* Record the first 2 entry directly to MTPT table */
>>> +A A A A A A A A A A A if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>> +A A A A A A A A A A A A A A A goto found;
>>> +A A A A A A A A A A A i++;
>>> +A A A A A A A }
>>> A A A A A }
>>>
>>> +found:
>>> A A A A A mpt_entry->pa0_l = cpu_to_le32(lower_32_bits(pages[0]));
>>> A A A A A roce_set_field(mpt_entry->byte_56_pa0_h, V2_MPT_BYTE_56_PA0_H_M,
>>> A A A A A A A A A A A A A A A A V2_MPT_BYTE_56_PA0_H_S,
>>> --
>>> 1.9.1
>>>
>
>
> _______________________________________________
> iommu mailing list
> iommu@lists.linux-foundation.org
> https://lists.linuxfoundation.org/mailman/listinfo/iommu
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-10-12 12:31 ` [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08 Wei Hu (Xavier)
2017-10-12 12:59 ` Robin Murphy
@ 2017-10-12 14:54 ` Leon Romanovsky
1 sibling, 0 replies; 14+ messages in thread
From: Leon Romanovsky @ 2017-10-12 14:54 UTC (permalink / raw)
To: Wei Hu (Xavier)
Cc: dledford, linux-rdma, lijun_nudt, oulijun, charles.chenxin,
liuyixian, linux-mm, zhangxiping3, xavier.huwei, linuxarm,
linux-kernel, shaobo.xu, shaoboxu, leizhen 00275356, joro, iommu
[-- Attachment #1: Type: text/plain, Size: 1451 bytes --]
On Thu, Oct 12, 2017 at 08:31:31PM +0800, Wei Hu (Xavier) wrote:
>
>
> On 2017/10/1 0:10, Leon Romanovsky wrote:
> > On Sat, Sep 30, 2017 at 05:28:59PM +0800, Wei Hu (Xavier) wrote:
> > > If the IOMMU is enabled, the length of sg obtained from
> > > __iommu_map_sg_attrs is not 4kB. When the IOVA is set with the sg
> > > dma address, the IOVA will not be page continuous. and the VA
> > > returned from dma_alloc_coherent is a vmalloc address. However,
> > > the VA obtained by the page_address is a discontinuous VA. Under
> > > these circumstances, the IOVA should be calculated based on the
> > > sg length, and record the VA returned from dma_alloc_coherent
> > > in the struct of hem.
> > >
> > > Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
> > > Signed-off-by: Shaobo Xu <xushaobo2@huawei.com>
> > > Signed-off-by: Lijun Ou <oulijun@huawei.com>
> > > ---
> > Doug,
> >
> > I didn't invest time in reviewing it, but having "is_vmalloc_addr" in
> > driver code to deal with dma_alloc_coherent is most probably wrong.
> >
> > Thanks
> Hi, Leon & Doug
> We refered the function named __ttm_dma_alloc_page in the kernel code as
> below:
> And there are similar methods in bch_bio_map and mem_to_page functions
> in current 4.14-rcx.
Let's put aside TTM, I don't know the rationale behind their implementation,
but both mem_to_page and bch_bio_map are don't operate on DMA addresses
and don't belong to HW driver code.
Thanks
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-10-12 12:59 ` Robin Murphy
@ 2017-11-01 7:46 ` Wei Hu (Xavier)
2017-11-01 12:26 ` Robin Murphy
0 siblings, 1 reply; 14+ messages in thread
From: Wei Hu (Xavier) @ 2017-11-01 7:46 UTC (permalink / raw)
To: Robin Murphy
Cc: Leon Romanovsky, shaobo.xu, xavier.huwei, lijun_nudt, oulijun,
linux-rdma, charles.chenxin, linuxarm, iommu, linux-kernel,
linux-mm, dledford, liuyixian, zhangxiping3, shaoboxu
On 2017/10/12 20:59, Robin Murphy wrote:
> On 12/10/17 13:31, Wei Hu (Xavier) wrote:
>>
>> On 2017/10/1 0:10, Leon Romanovsky wrote:
>>> On Sat, Sep 30, 2017 at 05:28:59PM +0800, Wei Hu (Xavier) wrote:
>>>> If the IOMMU is enabled, the length of sg obtained from
>>>> __iommu_map_sg_attrs is not 4kB. When the IOVA is set with the sg
>>>> dma address, the IOVA will not be page continuous. and the VA
>>>> returned from dma_alloc_coherent is a vmalloc address. However,
>>>> the VA obtained by the page_address is a discontinuous VA. Under
>>>> these circumstances, the IOVA should be calculated based on the
>>>> sg length, and record the VA returned from dma_alloc_coherent
>>>> in the struct of hem.
>>>>
>>>> Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
>>>> Signed-off-by: Shaobo Xu <xushaobo2@huawei.com>
>>>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>>>> ---
>>> Doug,
>>>
>>> I didn't invest time in reviewing it, but having "is_vmalloc_addr" in
>>> driver code to deal with dma_alloc_coherent is most probably wrong.
>>>
>>> Thanks
>> Hi, Leon & Doug
>> We refered the function named __ttm_dma_alloc_page in the kernel
>> code as below:
>> And there are similar methods in bch_bio_map and mem_to_page
>> functions in current 4.14-rcx.
>>
>> static struct dma_page *__ttm_dma_alloc_page(struct dma_pool *pool)
>> {
>> struct dma_page *d_page;
>>
>> d_page = kmalloc(sizeof(struct dma_page), GFP_KERNEL);
>> if (!d_page)
>> return NULL;
>>
>> d_page->vaddr = dma_alloc_coherent(pool->dev, pool->size,
>> &d_page->dma,
>> pool->gfp_flags);
>> if (d_page->vaddr) {
>> if (is_vmalloc_addr(d_page->vaddr))
>> d_page->p = vmalloc_to_page(d_page->vaddr);
>> else
>> d_page->p = virt_to_page(d_page->vaddr);
> There are cases on various architectures where neither of those is
> right. Whether those actually intersect with TTM or RDMA use-cases is
> another matter, of course.
>
> What definitely is a problem is if you ever take that page and end up
> accessing it through any virtual address other than the one explicitly
> returned by dma_alloc_coherent(). That can blow the coherency wide open
> and invite data loss, right up to killing the whole system with a
> machine check on certain architectures.
>
> Robin.
Hi, Robin
Thanks for your comment.
We have one problem and the related code as below.
1. call dma_alloc_coherent function serval times to alloc memory.
2. vmap the allocated memory pages.
3. software access memory by using the return virt addr of vmap
and hardware using the dma addr of dma_alloc_coherent.
When IOMMU is disabled in ARM64 architecture, we use virt_to_page()
before vmap(), it works. And when IOMMU is enabled using
virt_to_page() will cause calltrace later, we found the return
addr of dma_alloc_coherent is vmalloc addr, so we add the
condition judgement statement as below, it works.
for (i = 0; i < buf->nbufs; ++i)
pages[i] =
is_vmalloc_addr(buf->page_list[i].buf) ?
vmalloc_to_page(buf->page_list[i].buf) :
virt_to_page(buf->page_list[i].buf);
Can you give us suggestion? better method?
The related code as below:
buf->page_list = kcalloc(buf->nbufs, sizeof(*buf->page_list),
GFP_KERNEL);
if (!buf->page_list)
return -ENOMEM;
for (i = 0; i < buf->nbufs; ++i) {
buf->page_list[i].buf = dma_alloc_coherent(dev,
page_size, &t,
GFP_KERNEL);
if (!buf->page_list[i].buf)
goto err_free;
buf->page_list[i].map = t;
memset(buf->page_list[i].buf, 0, page_size);
}
pages = kmalloc_array(buf->nbufs, sizeof(*pages),
GFP_KERNEL);
if (!pages)
goto err_free;
for (i = 0; i < buf->nbufs; ++i)
pages[i] =
is_vmalloc_addr(buf->page_list[i].buf) ?
vmalloc_to_page(buf->page_list[i].buf) :
virt_to_page(buf->page_list[i].buf);
buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
PAGE_KERNEL);
kfree(pages);
if (!buf->direct.buf)
goto err_free;
Regards
Wei Hu
>> } else {
>> kfree(d_page);
>> d_page = NULL;
>> }
>> return d_page;
>> }
>>
>> Regards
>> Wei Hu
>>>> drivers/infiniband/hw/hns/hns_roce_alloc.c | 5 ++++-
>>>> drivers/infiniband/hw/hns/hns_roce_hem.c | 30
>>>> +++++++++++++++++++++++++++---
>>>> drivers/infiniband/hw/hns/hns_roce_hem.h | 6 ++++++
>>>> drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 22 +++++++++++++++-------
>>>> 4 files changed, 52 insertions(+), 11 deletions(-)
>>>>
>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>> b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>> index 3e4c525..a69cd4b 100644
>>>> --- a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>> @@ -243,7 +243,10 @@ int hns_roce_buf_alloc(struct hns_roce_dev
>>>> *hr_dev, u32 size, u32 max_direct,
>>>> goto err_free;
>>>>
>>>> for (i = 0; i < buf->nbufs; ++i)
>>>> - pages[i] = virt_to_page(buf->page_list[i].buf);
>>>> + pages[i] =
>>>> + is_vmalloc_addr(buf->page_list[i].buf) ?
>>>> + vmalloc_to_page(buf->page_list[i].buf) :
>>>> + virt_to_page(buf->page_list[i].buf);
>>>>
>>>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>>>> PAGE_KERNEL);
>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>> index 8388ae2..4a3d1d4 100644
>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>> @@ -200,6 +200,7 @@ static struct hns_roce_hem
>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>> gfp_t gfp_mask)
>>>> {
>>>> struct hns_roce_hem_chunk *chunk = NULL;
>>>> + struct hns_roce_vmalloc *vmalloc;
>>>> struct hns_roce_hem *hem;
>>>> struct scatterlist *mem;
>>>> int order;
>>>> @@ -227,6 +228,7 @@ static struct hns_roce_hem
>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>> sg_init_table(chunk->mem, HNS_ROCE_HEM_CHUNK_LEN);
>>>> chunk->npages = 0;
>>>> chunk->nsg = 0;
>>>> + memset(chunk->vmalloc, 0, sizeof(chunk->vmalloc));
>>>> list_add_tail(&chunk->list, &hem->chunk_list);
>>>> }
>>>>
>>>> @@ -243,7 +245,15 @@ static struct hns_roce_hem
>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>> if (!buf)
>>>> goto fail;
>>>>
>>>> - sg_set_buf(mem, buf, PAGE_SIZE << order);
>>>> + if (is_vmalloc_addr(buf)) {
>>>> + vmalloc = &chunk->vmalloc[chunk->npages];
>>>> + vmalloc->is_vmalloc_addr = true;
>>>> + vmalloc->vmalloc_addr = buf;
>>>> + sg_set_page(mem, vmalloc_to_page(buf),
>>>> + PAGE_SIZE << order, offset_in_page(buf));
>>>> + } else {
>>>> + sg_set_buf(mem, buf, PAGE_SIZE << order);
>>>> + }
>>>> WARN_ON(mem->offset);
>>>> sg_dma_len(mem) = PAGE_SIZE << order;
>>>>
>>>> @@ -262,17 +272,25 @@ static struct hns_roce_hem
>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>> void hns_roce_free_hem(struct hns_roce_dev *hr_dev, struct
>>>> hns_roce_hem *hem)
>>>> {
>>>> struct hns_roce_hem_chunk *chunk, *tmp;
>>>> + void *cpu_addr;
>>>> int i;
>>>>
>>>> if (!hem)
>>>> return;
>>>>
>>>> list_for_each_entry_safe(chunk, tmp, &hem->chunk_list, list) {
>>>> - for (i = 0; i < chunk->npages; ++i)
>>>> + for (i = 0; i < chunk->npages; ++i) {
>>>> + if (chunk->vmalloc[i].is_vmalloc_addr)
>>>> + cpu_addr = chunk->vmalloc[i].vmalloc_addr;
>>>> + else
>>>> + cpu_addr =
>>>> + lowmem_page_address(sg_page(&chunk->mem[i]));
>>>> +
>>>> dma_free_coherent(hr_dev->dev,
>>>> chunk->mem[i].length,
>>>> - lowmem_page_address(sg_page(&chunk->mem[i])),
>>>> + cpu_addr,
>>>> sg_dma_address(&chunk->mem[i]));
>>>> + }
>>>> kfree(chunk);
>>>> }
>>>>
>>>> @@ -774,6 +792,12 @@ void *hns_roce_table_find(struct hns_roce_dev
>>>> *hr_dev,
>>>>
>>>> if (chunk->mem[i].length > (u32)offset) {
>>>> page = sg_page(&chunk->mem[i]);
>>>> + if (chunk->vmalloc[i].is_vmalloc_addr) {
>>>> + mutex_unlock(&table->mutex);
>>>> + return page ?
>>>> + chunk->vmalloc[i].vmalloc_addr
>>>> + + offset : NULL;
>>>> + }
>>>> goto out;
>>>> }
>>>> offset -= chunk->mem[i].length;
>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>> index af28bbf..62d712a 100644
>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>> @@ -72,11 +72,17 @@ enum {
>>>> HNS_ROCE_HEM_PAGE_SIZE = 1 << HNS_ROCE_HEM_PAGE_SHIFT,
>>>> };
>>>>
>>>> +struct hns_roce_vmalloc {
>>>> + bool is_vmalloc_addr;
>>>> + void *vmalloc_addr;
>>>> +};
>>>> +
>>>> struct hns_roce_hem_chunk {
>>>> struct list_head list;
>>>> int npages;
>>>> int nsg;
>>>> struct scatterlist mem[HNS_ROCE_HEM_CHUNK_LEN];
>>>> + struct hns_roce_vmalloc vmalloc[HNS_ROCE_HEM_CHUNK_LEN];
>>>> };
>>>>
>>>> struct hns_roce_hem {
>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>> b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>> index b99d70a..9e19bf1 100644
>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>> @@ -1093,9 +1093,11 @@ static int hns_roce_v2_write_mtpt(void
>>>> *mb_buf, struct hns_roce_mr *mr,
>>>> {
>>>> struct hns_roce_v2_mpt_entry *mpt_entry;
>>>> struct scatterlist *sg;
>>>> + u64 page_addr = 0;
>>>> u64 *pages;
>>>> + int i = 0, j = 0;
>>>> + int len = 0;
>>>> int entry;
>>>> - int i;
>>>>
>>>> mpt_entry = mb_buf;
>>>> memset(mpt_entry, 0, sizeof(*mpt_entry));
>>>> @@ -1153,14 +1155,20 @@ static int hns_roce_v2_write_mtpt(void
>>>> *mb_buf, struct hns_roce_mr *mr,
>>>>
>>>> i = 0;
>>>> for_each_sg(mr->umem->sg_head.sgl, sg, mr->umem->nmap, entry) {
>>>> - pages[i] = ((u64)sg_dma_address(sg)) >> 6;
>>>> -
>>>> - /* Record the first 2 entry directly to MTPT table */
>>>> - if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>>> - break;
>>>> - i++;
>>>> + len = sg_dma_len(sg) >> PAGE_SHIFT;
>>>> + for (j = 0; j < len; ++j) {
>>>> + page_addr = sg_dma_address(sg) +
>>>> + (j << mr->umem->page_shift);
>>>> + pages[i] = page_addr >> 6;
>>>> +
>>>> + /* Record the first 2 entry directly to MTPT table */
>>>> + if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>>> + goto found;
>>>> + i++;
>>>> + }
>>>> }
>>>>
>>>> +found:
>>>> mpt_entry->pa0_l = cpu_to_le32(lower_32_bits(pages[0]));
>>>> roce_set_field(mpt_entry->byte_56_pa0_h, V2_MPT_BYTE_56_PA0_H_M,
>>>> V2_MPT_BYTE_56_PA0_H_S,
>>>> --
>>>> 1.9.1
>>>>
>>
>> _______________________________________________
>> iommu mailing list
>> iommu@lists.linux-foundation.org
>> https://lists.linuxfoundation.org/mailman/listinfo/iommu
>
> .
>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-11-01 7:46 ` Wei Hu (Xavier)
@ 2017-11-01 12:26 ` Robin Murphy
2017-11-07 2:45 ` Wei Hu (Xavier)
2017-11-09 1:36 ` Wei Hu (Xavier)
0 siblings, 2 replies; 14+ messages in thread
From: Robin Murphy @ 2017-11-01 12:26 UTC (permalink / raw)
To: Wei Hu (Xavier)
Cc: Leon Romanovsky, shaobo.xu, xavier.huwei, lijun_nudt, oulijun,
linux-rdma, charles.chenxin, linuxarm, iommu, linux-kernel,
linux-mm, dledford, liuyixian, zhangxiping3, shaoboxu
On 01/11/17 07:46, Wei Hu (Xavier) wrote:
>
>
> On 2017/10/12 20:59, Robin Murphy wrote:
>> On 12/10/17 13:31, Wei Hu (Xavier) wrote:
>>>
>>> On 2017/10/1 0:10, Leon Romanovsky wrote:
>>>> On Sat, Sep 30, 2017 at 05:28:59PM +0800, Wei Hu (Xavier) wrote:
>>>>> If the IOMMU is enabled, the length of sg obtained from
>>>>> __iommu_map_sg_attrs is not 4kB. When the IOVA is set with the sg
>>>>> dma address, the IOVA will not be page continuous. and the VA
>>>>> returned from dma_alloc_coherent is a vmalloc address. However,
>>>>> the VA obtained by the page_address is a discontinuous VA. Under
>>>>> these circumstances, the IOVA should be calculated based on the
>>>>> sg length, and record the VA returned from dma_alloc_coherent
>>>>> in the struct of hem.
>>>>>
>>>>> Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
>>>>> Signed-off-by: Shaobo Xu <xushaobo2@huawei.com>
>>>>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>>>>> ---
>>>> Doug,
>>>>
>>>> I didn't invest time in reviewing it, but having "is_vmalloc_addr" in
>>>> driver code to deal with dma_alloc_coherent is most probably wrong.
>>>>
>>>> Thanks
>>> Hi, Leon & Doug
>>> We refered the function named __ttm_dma_alloc_page in the kernel
>>> code as below:
>>> And there are similar methods in bch_bio_map and mem_to_page
>>> functions in current 4.14-rcx.
>>>
>>> static struct dma_page *__ttm_dma_alloc_page(struct dma_pool *pool)
>>> {
>>> struct dma_page *d_page;
>>>
>>> d_page = kmalloc(sizeof(struct dma_page), GFP_KERNEL);
>>> if (!d_page)
>>> return NULL;
>>>
>>> d_page->vaddr = dma_alloc_coherent(pool->dev, pool->size,
>>> &d_page->dma,
>>> pool->gfp_flags);
>>> if (d_page->vaddr) {
>>> if (is_vmalloc_addr(d_page->vaddr))
>>> d_page->p = vmalloc_to_page(d_page->vaddr);
>>> else
>>> d_page->p = virt_to_page(d_page->vaddr);
>> There are cases on various architectures where neither of those is
>> right. Whether those actually intersect with TTM or RDMA use-cases is
>> another matter, of course.
>>
>> What definitely is a problem is if you ever take that page and end up
>> accessing it through any virtual address other than the one explicitly
>> returned by dma_alloc_coherent(). That can blow the coherency wide open
>> and invite data loss, right up to killing the whole system with a
>> machine check on certain architectures.
>>
>> Robin.
> Hi, Robin
> Thanks for your comment.
>
> We have one problem and the related code as below.
> 1. call dma_alloc_coherent function serval times to alloc memory.
> 2. vmap the allocated memory pages.
> 3. software access memory by using the return virt addr of vmap
> and hardware using the dma addr of dma_alloc_coherent.
The simple answer is "don't do that". Seriously. dma_alloc_coherent()
gives you a CPU virtual address and a DMA address with which to access
your buffer, and that is the limit of what you may infer about it. You
have no guarantee that the virtual address is either in the linear map
or vmalloc, and not some other special place. You have no guarantee that
the underlying memory even has an associated struct page at all.
> When IOMMU is disabled in ARM64 architecture, we use virt_to_page()
> before vmap(), it works. And when IOMMU is enabled using
> virt_to_page() will cause calltrace later, we found the return
> addr of dma_alloc_coherent is vmalloc addr, so we add the
> condition judgement statement as below, it works.
> for (i = 0; i < buf->nbufs; ++i)
> pages[i] =
> is_vmalloc_addr(buf->page_list[i].buf) ?
> vmalloc_to_page(buf->page_list[i].buf) :
> virt_to_page(buf->page_list[i].buf);
> Can you give us suggestion? better method?
Oh my goodness, having now taken a closer look at this driver, I'm lost
for words in disbelief. To pick just one example:
u32 bits_per_long = BITS_PER_LONG;
...
if (bits_per_long == 64) {
/* memory mapping nonsense */
}
WTF does the size of a long have to do with DMA buffer management!?
Of course I can guess that it might be trying to make some tortuous
inference about vmalloc space being constrained on 32-bit platforms, but
still...
>
> The related code as below:
> buf->page_list = kcalloc(buf->nbufs, sizeof(*buf->page_list),
> GFP_KERNEL);
> if (!buf->page_list)
> return -ENOMEM;
>
> for (i = 0; i < buf->nbufs; ++i) {
> buf->page_list[i].buf = dma_alloc_coherent(dev,
> page_size, &t,
> GFP_KERNEL);
> if (!buf->page_list[i].buf)
> goto err_free;
>
> buf->page_list[i].map = t;
> memset(buf->page_list[i].buf, 0, page_size);
> }
>
> pages = kmalloc_array(buf->nbufs, sizeof(*pages),
> GFP_KERNEL);
> if (!pages)
> goto err_free;
>
> for (i = 0; i < buf->nbufs; ++i)
> pages[i] =
> is_vmalloc_addr(buf->page_list[i].buf) ?
> vmalloc_to_page(buf->page_list[i].buf) :
> virt_to_page(buf->page_list[i].buf);
>
> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
> PAGE_KERNEL);
> kfree(pages);
> if (!buf->direct.buf)
> goto err_free;
OK, this is complete crap. As above, you cannot assume that a struct
page even exists; even if it does you cannot assume that using a
PAGE_KERNEL mapping will not result in mismatched attributes,
unpredictable behaviour and data loss. Trying to remap coherent DMA
allocations like this is just egregiously wrong.
What I do like is that you can seemingly fix all this by simply deleting
hns_roce_buf::direct and all the garbage code related to it, and using
the page_list entries consistently because the alternate paths involving
those appear to do the right thing already.
That is, of course, assuming that the buffers involved can be so large
that it's not practical to just always make a single allocation and
fragment it into multiple descriptors if the hardware does have some
maximum length constraint - frankly I'm a little puzzled by the
PAGE_SIZE * 2 threshold, given that that's not a fixed size.
Robin.
>
> Regards
> Wei Hu
>>> } else {
>>> kfree(d_page);
>>> d_page = NULL;
>>> }
>>> return d_page;
>>> }
>>>
>>> Regards
>>> Wei Hu
>>>>> drivers/infiniband/hw/hns/hns_roce_alloc.c | 5 ++++-
>>>>> drivers/infiniband/hw/hns/hns_roce_hem.c | 30
>>>>> +++++++++++++++++++++++++++---
>>>>> drivers/infiniband/hw/hns/hns_roce_hem.h | 6 ++++++
>>>>> drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 22 +++++++++++++++-------
>>>>> 4 files changed, 52 insertions(+), 11 deletions(-)
>>>>>
>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>> b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>> index 3e4c525..a69cd4b 100644
>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>> @@ -243,7 +243,10 @@ int hns_roce_buf_alloc(struct hns_roce_dev
>>>>> *hr_dev, u32 size, u32 max_direct,
>>>>> goto err_free;
>>>>>
>>>>> for (i = 0; i < buf->nbufs; ++i)
>>>>> - pages[i] = virt_to_page(buf->page_list[i].buf);
>>>>> + pages[i] =
>>>>> + is_vmalloc_addr(buf->page_list[i].buf) ?
>>>>> + vmalloc_to_page(buf->page_list[i].buf) :
>>>>> + virt_to_page(buf->page_list[i].buf);
>>>>>
>>>>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>>>>> PAGE_KERNEL);
>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>> index 8388ae2..4a3d1d4 100644
>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>> @@ -200,6 +200,7 @@ static struct hns_roce_hem
>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>> gfp_t gfp_mask)
>>>>> {
>>>>> struct hns_roce_hem_chunk *chunk = NULL;
>>>>> + struct hns_roce_vmalloc *vmalloc;
>>>>> struct hns_roce_hem *hem;
>>>>> struct scatterlist *mem;
>>>>> int order;
>>>>> @@ -227,6 +228,7 @@ static struct hns_roce_hem
>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>> sg_init_table(chunk->mem, HNS_ROCE_HEM_CHUNK_LEN);
>>>>> chunk->npages = 0;
>>>>> chunk->nsg = 0;
>>>>> + memset(chunk->vmalloc, 0, sizeof(chunk->vmalloc));
>>>>> list_add_tail(&chunk->list, &hem->chunk_list);
>>>>> }
>>>>>
>>>>> @@ -243,7 +245,15 @@ static struct hns_roce_hem
>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>> if (!buf)
>>>>> goto fail;
>>>>>
>>>>> - sg_set_buf(mem, buf, PAGE_SIZE << order);
>>>>> + if (is_vmalloc_addr(buf)) {
>>>>> + vmalloc = &chunk->vmalloc[chunk->npages];
>>>>> + vmalloc->is_vmalloc_addr = true;
>>>>> + vmalloc->vmalloc_addr = buf;
>>>>> + sg_set_page(mem, vmalloc_to_page(buf),
>>>>> + PAGE_SIZE << order, offset_in_page(buf));
>>>>> + } else {
>>>>> + sg_set_buf(mem, buf, PAGE_SIZE << order);
>>>>> + }
>>>>> WARN_ON(mem->offset);
>>>>> sg_dma_len(mem) = PAGE_SIZE << order;
>>>>>
>>>>> @@ -262,17 +272,25 @@ static struct hns_roce_hem
>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>> void hns_roce_free_hem(struct hns_roce_dev *hr_dev, struct
>>>>> hns_roce_hem *hem)
>>>>> {
>>>>> struct hns_roce_hem_chunk *chunk, *tmp;
>>>>> + void *cpu_addr;
>>>>> int i;
>>>>>
>>>>> if (!hem)
>>>>> return;
>>>>>
>>>>> list_for_each_entry_safe(chunk, tmp, &hem->chunk_list, list) {
>>>>> - for (i = 0; i < chunk->npages; ++i)
>>>>> + for (i = 0; i < chunk->npages; ++i) {
>>>>> + if (chunk->vmalloc[i].is_vmalloc_addr)
>>>>> + cpu_addr = chunk->vmalloc[i].vmalloc_addr;
>>>>> + else
>>>>> + cpu_addr =
>>>>> + lowmem_page_address(sg_page(&chunk->mem[i]));
>>>>> +
>>>>> dma_free_coherent(hr_dev->dev,
>>>>> chunk->mem[i].length,
>>>>> - lowmem_page_address(sg_page(&chunk->mem[i])),
>>>>> + cpu_addr,
>>>>> sg_dma_address(&chunk->mem[i]));
>>>>> + }
>>>>> kfree(chunk);
>>>>> }
>>>>>
>>>>> @@ -774,6 +792,12 @@ void *hns_roce_table_find(struct hns_roce_dev
>>>>> *hr_dev,
>>>>>
>>>>> if (chunk->mem[i].length > (u32)offset) {
>>>>> page = sg_page(&chunk->mem[i]);
>>>>> + if (chunk->vmalloc[i].is_vmalloc_addr) {
>>>>> + mutex_unlock(&table->mutex);
>>>>> + return page ?
>>>>> + chunk->vmalloc[i].vmalloc_addr
>>>>> + + offset : NULL;
>>>>> + }
>>>>> goto out;
>>>>> }
>>>>> offset -= chunk->mem[i].length;
>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>> index af28bbf..62d712a 100644
>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>> @@ -72,11 +72,17 @@ enum {
>>>>> HNS_ROCE_HEM_PAGE_SIZE = 1 << HNS_ROCE_HEM_PAGE_SHIFT,
>>>>> };
>>>>>
>>>>> +struct hns_roce_vmalloc {
>>>>> + bool is_vmalloc_addr;
>>>>> + void *vmalloc_addr;
>>>>> +};
>>>>> +
>>>>> struct hns_roce_hem_chunk {
>>>>> struct list_head list;
>>>>> int npages;
>>>>> int nsg;
>>>>> struct scatterlist mem[HNS_ROCE_HEM_CHUNK_LEN];
>>>>> + struct hns_roce_vmalloc vmalloc[HNS_ROCE_HEM_CHUNK_LEN];
>>>>> };
>>>>>
>>>>> struct hns_roce_hem {
>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>> b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>> index b99d70a..9e19bf1 100644
>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>> @@ -1093,9 +1093,11 @@ static int hns_roce_v2_write_mtpt(void
>>>>> *mb_buf, struct hns_roce_mr *mr,
>>>>> {
>>>>> struct hns_roce_v2_mpt_entry *mpt_entry;
>>>>> struct scatterlist *sg;
>>>>> + u64 page_addr = 0;
>>>>> u64 *pages;
>>>>> + int i = 0, j = 0;
>>>>> + int len = 0;
>>>>> int entry;
>>>>> - int i;
>>>>>
>>>>> mpt_entry = mb_buf;
>>>>> memset(mpt_entry, 0, sizeof(*mpt_entry));
>>>>> @@ -1153,14 +1155,20 @@ static int hns_roce_v2_write_mtpt(void
>>>>> *mb_buf, struct hns_roce_mr *mr,
>>>>>
>>>>> i = 0;
>>>>> for_each_sg(mr->umem->sg_head.sgl, sg, mr->umem->nmap, entry) {
>>>>> - pages[i] = ((u64)sg_dma_address(sg)) >> 6;
>>>>> -
>>>>> - /* Record the first 2 entry directly to MTPT table */
>>>>> - if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>>>> - break;
>>>>> - i++;
>>>>> + len = sg_dma_len(sg) >> PAGE_SHIFT;
>>>>> + for (j = 0; j < len; ++j) {
>>>>> + page_addr = sg_dma_address(sg) +
>>>>> + (j << mr->umem->page_shift);
>>>>> + pages[i] = page_addr >> 6;
>>>>> +
>>>>> + /* Record the first 2 entry directly to MTPT table */
>>>>> + if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>>>> + goto found;
>>>>> + i++;
>>>>> + }
>>>>> }
>>>>>
>>>>> +found:
>>>>> mpt_entry->pa0_l = cpu_to_le32(lower_32_bits(pages[0]));
>>>>> roce_set_field(mpt_entry->byte_56_pa0_h, V2_MPT_BYTE_56_PA0_H_M,
>>>>> V2_MPT_BYTE_56_PA0_H_S,
>>>>> --
>>>>> 1.9.1
>>>>>
>>>
>>> _______________________________________________
>>> iommu mailing list
>>> iommu@lists.linux-foundation.org
>>> https://lists.linuxfoundation.org/mailman/listinfo/iommu
>>
>> .
>>
>
>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-11-01 12:26 ` Robin Murphy
@ 2017-11-07 2:45 ` Wei Hu (Xavier)
2017-11-07 6:32 ` Leon Romanovsky
2017-11-07 15:48 ` Jason Gunthorpe
2017-11-09 1:36 ` Wei Hu (Xavier)
1 sibling, 2 replies; 14+ messages in thread
From: Wei Hu (Xavier) @ 2017-11-07 2:45 UTC (permalink / raw)
To: Robin Murphy
Cc: Leon Romanovsky, shaobo.xu, xavier.huwei, lijun_nudt, oulijun,
linux-rdma, charles.chenxin, linuxarm, iommu, linux-kernel,
linux-mm, dledford, liuyixian, zhangxiping3, shaoboxu
On 2017/11/1 20:26, Robin Murphy wrote:
> On 01/11/17 07:46, Wei Hu (Xavier) wrote:
>>
>> On 2017/10/12 20:59, Robin Murphy wrote:
>>> On 12/10/17 13:31, Wei Hu (Xavier) wrote:
>>>> On 2017/10/1 0:10, Leon Romanovsky wrote:
>>>>> On Sat, Sep 30, 2017 at 05:28:59PM +0800, Wei Hu (Xavier) wrote:
>>>>>> If the IOMMU is enabled, the length of sg obtained from
>>>>>> __iommu_map_sg_attrs is not 4kB. When the IOVA is set with the sg
>>>>>> dma address, the IOVA will not be page continuous. and the VA
>>>>>> returned from dma_alloc_coherent is a vmalloc address. However,
>>>>>> the VA obtained by the page_address is a discontinuous VA. Under
>>>>>> these circumstances, the IOVA should be calculated based on the
>>>>>> sg length, and record the VA returned from dma_alloc_coherent
>>>>>> in the struct of hem.
>>>>>>
>>>>>> Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
>>>>>> Signed-off-by: Shaobo Xu <xushaobo2@huawei.com>
>>>>>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>>>>>> ---
>>>>> Doug,
>>>>>
>>>>> I didn't invest time in reviewing it, but having "is_vmalloc_addr" in
>>>>> driver code to deal with dma_alloc_coherent is most probably wrong.
>>>>>
>>>>> Thanks
>>>> Hi, Leon & Doug
>>>> We refered the function named __ttm_dma_alloc_page in the kernel
>>>> code as below:
>>>> And there are similar methods in bch_bio_map and mem_to_page
>>>> functions in current 4.14-rcx.
>>>>
>>>> static struct dma_page *__ttm_dma_alloc_page(struct dma_pool *pool)
>>>> {
>>>> struct dma_page *d_page;
>>>>
>>>> d_page = kmalloc(sizeof(struct dma_page), GFP_KERNEL);
>>>> if (!d_page)
>>>> return NULL;
>>>>
>>>> d_page->vaddr = dma_alloc_coherent(pool->dev, pool->size,
>>>> &d_page->dma,
>>>> pool->gfp_flags);
>>>> if (d_page->vaddr) {
>>>> if (is_vmalloc_addr(d_page->vaddr))
>>>> d_page->p = vmalloc_to_page(d_page->vaddr);
>>>> else
>>>> d_page->p = virt_to_page(d_page->vaddr);
>>> There are cases on various architectures where neither of those is
>>> right. Whether those actually intersect with TTM or RDMA use-cases is
>>> another matter, of course.
>>>
>>> What definitely is a problem is if you ever take that page and end up
>>> accessing it through any virtual address other than the one explicitly
>>> returned by dma_alloc_coherent(). That can blow the coherency wide open
>>> and invite data loss, right up to killing the whole system with a
>>> machine check on certain architectures.
>>>
>>> Robin.
>> Hi, Robin
>> Thanks for your comment.
>>
>> We have one problem and the related code as below.
>> 1. call dma_alloc_coherent function serval times to alloc memory.
>> 2. vmap the allocated memory pages.
>> 3. software access memory by using the return virt addr of vmap
>> and hardware using the dma addr of dma_alloc_coherent.
> The simple answer is "don't do that". Seriously. dma_alloc_coherent()
> gives you a CPU virtual address and a DMA address with which to access
> your buffer, and that is the limit of what you may infer about it. You
> have no guarantee that the virtual address is either in the linear map
> or vmalloc, and not some other special place. You have no guarantee that
> the underlying memory even has an associated struct page at all.
>
>> When IOMMU is disabled in ARM64 architecture, we use virt_to_page()
>> before vmap(), it works. And when IOMMU is enabled using
>> virt_to_page() will cause calltrace later, we found the return
>> addr of dma_alloc_coherent is vmalloc addr, so we add the
>> condition judgement statement as below, it works.
>> for (i = 0; i < buf->nbufs; ++i)
>> pages[i] =
>> is_vmalloc_addr(buf->page_list[i].buf) ?
>> vmalloc_to_page(buf->page_list[i].buf) :
>> virt_to_page(buf->page_list[i].buf);
>> Can you give us suggestion? better method?
> Oh my goodness, having now taken a closer look at this driver, I'm lost
> for words in disbelief. To pick just one example:
>
> u32 bits_per_long = BITS_PER_LONG;
> ...
> if (bits_per_long == 64) {
> /* memory mapping nonsense */
> }
>
> WTF does the size of a long have to do with DMA buffer management!?
>
> Of course I can guess that it might be trying to make some tortuous
> inference about vmalloc space being constrained on 32-bit platforms, but
> still...
>
>> The related code as below:
>> buf->page_list = kcalloc(buf->nbufs, sizeof(*buf->page_list),
>> GFP_KERNEL);
>> if (!buf->page_list)
>> return -ENOMEM;
>>
>> for (i = 0; i < buf->nbufs; ++i) {
>> buf->page_list[i].buf = dma_alloc_coherent(dev,
>> page_size, &t,
>> GFP_KERNEL);
>> if (!buf->page_list[i].buf)
>> goto err_free;
>>
>> buf->page_list[i].map = t;
>> memset(buf->page_list[i].buf, 0, page_size);
>> }
>>
>> pages = kmalloc_array(buf->nbufs, sizeof(*pages),
>> GFP_KERNEL);
>> if (!pages)
>> goto err_free;
>>
>> for (i = 0; i < buf->nbufs; ++i)
>> pages[i] =
>> is_vmalloc_addr(buf->page_list[i].buf) ?
>> vmalloc_to_page(buf->page_list[i].buf) :
>> virt_to_page(buf->page_list[i].buf);
>>
>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>> PAGE_KERNEL);
>> kfree(pages);
>> if (!buf->direct.buf)
>> goto err_free;
> OK, this is complete crap. As above, you cannot assume that a struct
> page even exists; even if it does you cannot assume that using a
> PAGE_KERNEL mapping will not result in mismatched attributes,
> unpredictable behaviour and data loss. Trying to remap coherent DMA
> allocations like this is just egregiously wrong.
>
> What I do like is that you can seemingly fix all this by simply deleting
> hns_roce_buf::direct and all the garbage code related to it, and using
> the page_list entries consistently because the alternate paths involving
> those appear to do the right thing already.
>
> That is, of course, assuming that the buffers involved can be so large
> that it's not practical to just always make a single allocation and
> fragment it into multiple descriptors if the hardware does have some
> maximum length constraint - frankly I'm a little puzzled by the
> PAGE_SIZE * 2 threshold, given that that's not a fixed size.
>
> Robin.
Hii 1/4 ?Robin
We reconstruct the code as below:
It replaces dma_alloc_coherent with __get_free_pages and
dma_map_single
functions. So, we can vmap serveral ptrs returned by
__get_free_pages, right?
buf->page_list = kcalloc(buf->nbufs, sizeof(*buf->page_list),
GFP_KERNEL);
if (!buf->page_list)
return -ENOMEM;
for (i = 0; i < buf->nbufs; ++i) {
ptr = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
get_order(page_size));
if (!ptr) {
dev_err(dev, "Alloc pages error.\n");
goto err_free;
}
t = dma_map_single(dev, ptr, page_size,
DMA_BIDIRECTIONAL);
if (dma_mapping_error(dev, t)) {
dev_err(dev, "DMA mapping error.\n");
free_pages((unsigned long)ptr,
get_order(page_size));
goto err_free;
}
buf->page_list[i].buf = ptr;
buf->page_list[i].map = t;
}
pages = kmalloc_array(buf->nbufs, sizeof(*pages),
GFP_KERNEL);
if (!pages)
goto err_free;
for (i = 0; i < buf->nbufs; ++i)
pages[i] = virt_to_page(buf->page_list[i].buf);
buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
PAGE_KERNEL);
kfree(pages);
if (!buf->direct.buf)
goto err_free;
Regards
Wei Hu
>> Regards
>> Wei Hu
>>>> } else {
>>>> kfree(d_page);
>>>> d_page = NULL;
>>>> }
>>>> return d_page;
>>>> }
>>>>
>>>> Regards
>>>> Wei Hu
>>>>>> drivers/infiniband/hw/hns/hns_roce_alloc.c | 5 ++++-
>>>>>> drivers/infiniband/hw/hns/hns_roce_hem.c | 30
>>>>>> +++++++++++++++++++++++++++---
>>>>>> drivers/infiniband/hw/hns/hns_roce_hem.h | 6 ++++++
>>>>>> drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 22 +++++++++++++++-------
>>>>>> 4 files changed, 52 insertions(+), 11 deletions(-)
>>>>>>
>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>> b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>> index 3e4c525..a69cd4b 100644
>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>> @@ -243,7 +243,10 @@ int hns_roce_buf_alloc(struct hns_roce_dev
>>>>>> *hr_dev, u32 size, u32 max_direct,
>>>>>> goto err_free;
>>>>>>
>>>>>> for (i = 0; i < buf->nbufs; ++i)
>>>>>> - pages[i] = virt_to_page(buf->page_list[i].buf);
>>>>>> + pages[i] =
>>>>>> + is_vmalloc_addr(buf->page_list[i].buf) ?
>>>>>> + vmalloc_to_page(buf->page_list[i].buf) :
>>>>>> + virt_to_page(buf->page_list[i].buf);
>>>>>>
>>>>>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>>>>>> PAGE_KERNEL);
>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>> index 8388ae2..4a3d1d4 100644
>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>> @@ -200,6 +200,7 @@ static struct hns_roce_hem
>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>> gfp_t gfp_mask)
>>>>>> {
>>>>>> struct hns_roce_hem_chunk *chunk = NULL;
>>>>>> + struct hns_roce_vmalloc *vmalloc;
>>>>>> struct hns_roce_hem *hem;
>>>>>> struct scatterlist *mem;
>>>>>> int order;
>>>>>> @@ -227,6 +228,7 @@ static struct hns_roce_hem
>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>> sg_init_table(chunk->mem, HNS_ROCE_HEM_CHUNK_LEN);
>>>>>> chunk->npages = 0;
>>>>>> chunk->nsg = 0;
>>>>>> + memset(chunk->vmalloc, 0, sizeof(chunk->vmalloc));
>>>>>> list_add_tail(&chunk->list, &hem->chunk_list);
>>>>>> }
>>>>>>
>>>>>> @@ -243,7 +245,15 @@ static struct hns_roce_hem
>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>> if (!buf)
>>>>>> goto fail;
>>>>>>
>>>>>> - sg_set_buf(mem, buf, PAGE_SIZE << order);
>>>>>> + if (is_vmalloc_addr(buf)) {
>>>>>> + vmalloc = &chunk->vmalloc[chunk->npages];
>>>>>> + vmalloc->is_vmalloc_addr = true;
>>>>>> + vmalloc->vmalloc_addr = buf;
>>>>>> + sg_set_page(mem, vmalloc_to_page(buf),
>>>>>> + PAGE_SIZE << order, offset_in_page(buf));
>>>>>> + } else {
>>>>>> + sg_set_buf(mem, buf, PAGE_SIZE << order);
>>>>>> + }
>>>>>> WARN_ON(mem->offset);
>>>>>> sg_dma_len(mem) = PAGE_SIZE << order;
>>>>>>
>>>>>> @@ -262,17 +272,25 @@ static struct hns_roce_hem
>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>> void hns_roce_free_hem(struct hns_roce_dev *hr_dev, struct
>>>>>> hns_roce_hem *hem)
>>>>>> {
>>>>>> struct hns_roce_hem_chunk *chunk, *tmp;
>>>>>> + void *cpu_addr;
>>>>>> int i;
>>>>>>
>>>>>> if (!hem)
>>>>>> return;
>>>>>>
>>>>>> list_for_each_entry_safe(chunk, tmp, &hem->chunk_list, list) {
>>>>>> - for (i = 0; i < chunk->npages; ++i)
>>>>>> + for (i = 0; i < chunk->npages; ++i) {
>>>>>> + if (chunk->vmalloc[i].is_vmalloc_addr)
>>>>>> + cpu_addr = chunk->vmalloc[i].vmalloc_addr;
>>>>>> + else
>>>>>> + cpu_addr =
>>>>>> + lowmem_page_address(sg_page(&chunk->mem[i]));
>>>>>> +
>>>>>> dma_free_coherent(hr_dev->dev,
>>>>>> chunk->mem[i].length,
>>>>>> - lowmem_page_address(sg_page(&chunk->mem[i])),
>>>>>> + cpu_addr,
>>>>>> sg_dma_address(&chunk->mem[i]));
>>>>>> + }
>>>>>> kfree(chunk);
>>>>>> }
>>>>>>
>>>>>> @@ -774,6 +792,12 @@ void *hns_roce_table_find(struct hns_roce_dev
>>>>>> *hr_dev,
>>>>>>
>>>>>> if (chunk->mem[i].length > (u32)offset) {
>>>>>> page = sg_page(&chunk->mem[i]);
>>>>>> + if (chunk->vmalloc[i].is_vmalloc_addr) {
>>>>>> + mutex_unlock(&table->mutex);
>>>>>> + return page ?
>>>>>> + chunk->vmalloc[i].vmalloc_addr
>>>>>> + + offset : NULL;
>>>>>> + }
>>>>>> goto out;
>>>>>> }
>>>>>> offset -= chunk->mem[i].length;
>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>> index af28bbf..62d712a 100644
>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>> @@ -72,11 +72,17 @@ enum {
>>>>>> HNS_ROCE_HEM_PAGE_SIZE = 1 << HNS_ROCE_HEM_PAGE_SHIFT,
>>>>>> };
>>>>>>
>>>>>> +struct hns_roce_vmalloc {
>>>>>> + bool is_vmalloc_addr;
>>>>>> + void *vmalloc_addr;
>>>>>> +};
>>>>>> +
>>>>>> struct hns_roce_hem_chunk {
>>>>>> struct list_head list;
>>>>>> int npages;
>>>>>> int nsg;
>>>>>> struct scatterlist mem[HNS_ROCE_HEM_CHUNK_LEN];
>>>>>> + struct hns_roce_vmalloc vmalloc[HNS_ROCE_HEM_CHUNK_LEN];
>>>>>> };
>>>>>>
>>>>>> struct hns_roce_hem {
>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>> b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>> index b99d70a..9e19bf1 100644
>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>> @@ -1093,9 +1093,11 @@ static int hns_roce_v2_write_mtpt(void
>>>>>> *mb_buf, struct hns_roce_mr *mr,
>>>>>> {
>>>>>> struct hns_roce_v2_mpt_entry *mpt_entry;
>>>>>> struct scatterlist *sg;
>>>>>> + u64 page_addr = 0;
>>>>>> u64 *pages;
>>>>>> + int i = 0, j = 0;
>>>>>> + int len = 0;
>>>>>> int entry;
>>>>>> - int i;
>>>>>>
>>>>>> mpt_entry = mb_buf;
>>>>>> memset(mpt_entry, 0, sizeof(*mpt_entry));
>>>>>> @@ -1153,14 +1155,20 @@ static int hns_roce_v2_write_mtpt(void
>>>>>> *mb_buf, struct hns_roce_mr *mr,
>>>>>>
>>>>>> i = 0;
>>>>>> for_each_sg(mr->umem->sg_head.sgl, sg, mr->umem->nmap, entry) {
>>>>>> - pages[i] = ((u64)sg_dma_address(sg)) >> 6;
>>>>>> -
>>>>>> - /* Record the first 2 entry directly to MTPT table */
>>>>>> - if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>>>>> - break;
>>>>>> - i++;
>>>>>> + len = sg_dma_len(sg) >> PAGE_SHIFT;
>>>>>> + for (j = 0; j < len; ++j) {
>>>>>> + page_addr = sg_dma_address(sg) +
>>>>>> + (j << mr->umem->page_shift);
>>>>>> + pages[i] = page_addr >> 6;
>>>>>> +
>>>>>> + /* Record the first 2 entry directly to MTPT table */
>>>>>> + if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>>>>> + goto found;
>>>>>> + i++;
>>>>>> + }
>>>>>> }
>>>>>>
>>>>>> +found:
>>>>>> mpt_entry->pa0_l = cpu_to_le32(lower_32_bits(pages[0]));
>>>>>> roce_set_field(mpt_entry->byte_56_pa0_h, V2_MPT_BYTE_56_PA0_H_M,
>>>>>> V2_MPT_BYTE_56_PA0_H_S,
>>>>>> --
>>>>>> 1.9.1
>>>>>>
>>>> _______________________________________________
>>>> iommu mailing list
>>>> iommu@lists.linux-foundation.org
>>>> https://lists.linuxfoundation.org/mailman/listinfo/iommu
>>> .
>>>
>>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
> .
>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-11-07 2:45 ` Wei Hu (Xavier)
@ 2017-11-07 6:32 ` Leon Romanovsky
2017-11-09 1:17 ` Wei Hu (Xavier)
2017-11-07 15:48 ` Jason Gunthorpe
1 sibling, 1 reply; 14+ messages in thread
From: Leon Romanovsky @ 2017-11-07 6:32 UTC (permalink / raw)
To: Wei Hu (Xavier)
Cc: Robin Murphy, shaobo.xu, xavier.huwei, lijun_nudt, oulijun,
linux-rdma, charles.chenxin, linuxarm, iommu, linux-kernel,
linux-mm, dledford, liuyixian, zhangxiping3, shaoboxu
[-- Attachment #1: Type: text/plain, Size: 18038 bytes --]
On Tue, Nov 07, 2017 at 10:45:29AM +0800, Wei Hu (Xavier) wrote:
>
>
> On 2017/11/1 20:26, Robin Murphy wrote:
> > On 01/11/17 07:46, Wei Hu (Xavier) wrote:
> >>
> >> On 2017/10/12 20:59, Robin Murphy wrote:
> >>> On 12/10/17 13:31, Wei Hu (Xavier) wrote:
> >>>> On 2017/10/1 0:10, Leon Romanovsky wrote:
> >>>>> On Sat, Sep 30, 2017 at 05:28:59PM +0800, Wei Hu (Xavier) wrote:
> >>>>>> If the IOMMU is enabled, the length of sg obtained from
> >>>>>> __iommu_map_sg_attrs is not 4kB. When the IOVA is set with the sg
> >>>>>> dma address, the IOVA will not be page continuous. and the VA
> >>>>>> returned from dma_alloc_coherent is a vmalloc address. However,
> >>>>>> the VA obtained by the page_address is a discontinuous VA. Under
> >>>>>> these circumstances, the IOVA should be calculated based on the
> >>>>>> sg length, and record the VA returned from dma_alloc_coherent
> >>>>>> in the struct of hem.
> >>>>>>
> >>>>>> Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
> >>>>>> Signed-off-by: Shaobo Xu <xushaobo2@huawei.com>
> >>>>>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
> >>>>>> ---
> >>>>> Doug,
> >>>>>
> >>>>> I didn't invest time in reviewing it, but having "is_vmalloc_addr" in
> >>>>> driver code to deal with dma_alloc_coherent is most probably wrong.
> >>>>>
> >>>>> Thanks
> >>>> Hi, Leon & Doug
> >>>> We refered the function named __ttm_dma_alloc_page in the kernel
> >>>> code as below:
> >>>> And there are similar methods in bch_bio_map and mem_to_page
> >>>> functions in current 4.14-rcx.
> >>>>
> >>>> static struct dma_page *__ttm_dma_alloc_page(struct dma_pool *pool)
> >>>> {
> >>>> struct dma_page *d_page;
> >>>>
> >>>> d_page = kmalloc(sizeof(struct dma_page), GFP_KERNEL);
> >>>> if (!d_page)
> >>>> return NULL;
> >>>>
> >>>> d_page->vaddr = dma_alloc_coherent(pool->dev, pool->size,
> >>>> &d_page->dma,
> >>>> pool->gfp_flags);
> >>>> if (d_page->vaddr) {
> >>>> if (is_vmalloc_addr(d_page->vaddr))
> >>>> d_page->p = vmalloc_to_page(d_page->vaddr);
> >>>> else
> >>>> d_page->p = virt_to_page(d_page->vaddr);
> >>> There are cases on various architectures where neither of those is
> >>> right. Whether those actually intersect with TTM or RDMA use-cases is
> >>> another matter, of course.
> >>>
> >>> What definitely is a problem is if you ever take that page and end up
> >>> accessing it through any virtual address other than the one explicitly
> >>> returned by dma_alloc_coherent(). That can blow the coherency wide open
> >>> and invite data loss, right up to killing the whole system with a
> >>> machine check on certain architectures.
> >>>
> >>> Robin.
> >> Hi, Robin
> >> Thanks for your comment.
> >>
> >> We have one problem and the related code as below.
> >> 1. call dma_alloc_coherent function serval times to alloc memory.
> >> 2. vmap the allocated memory pages.
> >> 3. software access memory by using the return virt addr of vmap
> >> and hardware using the dma addr of dma_alloc_coherent.
> > The simple answer is "don't do that". Seriously. dma_alloc_coherent()
> > gives you a CPU virtual address and a DMA address with which to access
> > your buffer, and that is the limit of what you may infer about it. You
> > have no guarantee that the virtual address is either in the linear map
> > or vmalloc, and not some other special place. You have no guarantee that
> > the underlying memory even has an associated struct page at all.
> >
> >> When IOMMU is disabled in ARM64 architecture, we use virt_to_page()
> >> before vmap(), it works. And when IOMMU is enabled using
> >> virt_to_page() will cause calltrace later, we found the return
> >> addr of dma_alloc_coherent is vmalloc addr, so we add the
> >> condition judgement statement as below, it works.
> >> for (i = 0; i < buf->nbufs; ++i)
> >> pages[i] =
> >> is_vmalloc_addr(buf->page_list[i].buf) ?
> >> vmalloc_to_page(buf->page_list[i].buf) :
> >> virt_to_page(buf->page_list[i].buf);
> >> Can you give us suggestion? better method?
> > Oh my goodness, having now taken a closer look at this driver, I'm lost
> > for words in disbelief. To pick just one example:
> >
> > u32 bits_per_long = BITS_PER_LONG;
> > ...
> > if (bits_per_long == 64) {
> > /* memory mapping nonsense */
> > }
> >
> > WTF does the size of a long have to do with DMA buffer management!?
> >
> > Of course I can guess that it might be trying to make some tortuous
> > inference about vmalloc space being constrained on 32-bit platforms, but
> > still...
> >
> >> The related code as below:
> >> buf->page_list = kcalloc(buf->nbufs, sizeof(*buf->page_list),
> >> GFP_KERNEL);
> >> if (!buf->page_list)
> >> return -ENOMEM;
> >>
> >> for (i = 0; i < buf->nbufs; ++i) {
> >> buf->page_list[i].buf = dma_alloc_coherent(dev,
> >> page_size, &t,
> >> GFP_KERNEL);
> >> if (!buf->page_list[i].buf)
> >> goto err_free;
> >>
> >> buf->page_list[i].map = t;
> >> memset(buf->page_list[i].buf, 0, page_size);
> >> }
> >>
> >> pages = kmalloc_array(buf->nbufs, sizeof(*pages),
> >> GFP_KERNEL);
> >> if (!pages)
> >> goto err_free;
> >>
> >> for (i = 0; i < buf->nbufs; ++i)
> >> pages[i] =
> >> is_vmalloc_addr(buf->page_list[i].buf) ?
> >> vmalloc_to_page(buf->page_list[i].buf) :
> >> virt_to_page(buf->page_list[i].buf);
> >>
> >> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
> >> PAGE_KERNEL);
> >> kfree(pages);
> >> if (!buf->direct.buf)
> >> goto err_free;
> > OK, this is complete crap. As above, you cannot assume that a struct
> > page even exists; even if it does you cannot assume that using a
> > PAGE_KERNEL mapping will not result in mismatched attributes,
> > unpredictable behaviour and data loss. Trying to remap coherent DMA
> > allocations like this is just egregiously wrong.
> >
> > What I do like is that you can seemingly fix all this by simply deleting
> > hns_roce_buf::direct and all the garbage code related to it, and using
> > the page_list entries consistently because the alternate paths involving
> > those appear to do the right thing already.
> >
> > That is, of course, assuming that the buffers involved can be so large
> > that it's not practical to just always make a single allocation and
> > fragment it into multiple descriptors if the hardware does have some
> > maximum length constraint - frankly I'm a little puzzled by the
> > PAGE_SIZE * 2 threshold, given that that's not a fixed size.
> >
> > Robin.
> Hi，Robin
>
> We reconstruct the code as below:
> It replaces dma_alloc_coherent with __get_free_pages and
> dma_map_single
> functions. So, we can vmap serveral ptrs returned by
> __get_free_pages, right?
Most probably not, you should get rid of your virt_to_page/vmap calls.
Thanks
>
>
> buf->page_list = kcalloc(buf->nbufs, sizeof(*buf->page_list),
> GFP_KERNEL);
> if (!buf->page_list)
> return -ENOMEM;
>
> for (i = 0; i < buf->nbufs; ++i) {
> ptr = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
> get_order(page_size));
> if (!ptr) {
> dev_err(dev, "Alloc pages error.\n");
> goto err_free;
> }
>
> t = dma_map_single(dev, ptr, page_size,
> DMA_BIDIRECTIONAL);
> if (dma_mapping_error(dev, t)) {
> dev_err(dev, "DMA mapping error.\n");
> free_pages((unsigned long)ptr,
> get_order(page_size));
> goto err_free;
> }
>
> buf->page_list[i].buf = ptr;
> buf->page_list[i].map = t;
> }
>
> pages = kmalloc_array(buf->nbufs, sizeof(*pages),
> GFP_KERNEL);
> if (!pages)
> goto err_free;
>
> for (i = 0; i < buf->nbufs; ++i)
> pages[i] = virt_to_page(buf->page_list[i].buf);
>
> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
> PAGE_KERNEL);
> kfree(pages);
> if (!buf->direct.buf)
> goto err_free;
>
>
> Regards
> Wei Hu
> >> Regards
> >> Wei Hu
> >>>> } else {
> >>>> kfree(d_page);
> >>>> d_page = NULL;
> >>>> }
> >>>> return d_page;
> >>>> }
> >>>>
> >>>> Regards
> >>>> Wei Hu
> >>>>>> drivers/infiniband/hw/hns/hns_roce_alloc.c | 5 ++++-
> >>>>>> drivers/infiniband/hw/hns/hns_roce_hem.c | 30
> >>>>>> +++++++++++++++++++++++++++---
> >>>>>> drivers/infiniband/hw/hns/hns_roce_hem.h | 6 ++++++
> >>>>>> drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 22 +++++++++++++++-------
> >>>>>> 4 files changed, 52 insertions(+), 11 deletions(-)
> >>>>>>
> >>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c
> >>>>>> b/drivers/infiniband/hw/hns/hns_roce_alloc.c
> >>>>>> index 3e4c525..a69cd4b 100644
> >>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_alloc.c
> >>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c
> >>>>>> @@ -243,7 +243,10 @@ int hns_roce_buf_alloc(struct hns_roce_dev
> >>>>>> *hr_dev, u32 size, u32 max_direct,
> >>>>>> goto err_free;
> >>>>>>
> >>>>>> for (i = 0; i < buf->nbufs; ++i)
> >>>>>> - pages[i] = virt_to_page(buf->page_list[i].buf);
> >>>>>> + pages[i] =
> >>>>>> + is_vmalloc_addr(buf->page_list[i].buf) ?
> >>>>>> + vmalloc_to_page(buf->page_list[i].buf) :
> >>>>>> + virt_to_page(buf->page_list[i].buf);
> >>>>>>
> >>>>>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
> >>>>>> PAGE_KERNEL);
> >>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c
> >>>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.c
> >>>>>> index 8388ae2..4a3d1d4 100644
> >>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.c
> >>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c
> >>>>>> @@ -200,6 +200,7 @@ static struct hns_roce_hem
> >>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
> >>>>>> gfp_t gfp_mask)
> >>>>>> {
> >>>>>> struct hns_roce_hem_chunk *chunk = NULL;
> >>>>>> + struct hns_roce_vmalloc *vmalloc;
> >>>>>> struct hns_roce_hem *hem;
> >>>>>> struct scatterlist *mem;
> >>>>>> int order;
> >>>>>> @@ -227,6 +228,7 @@ static struct hns_roce_hem
> >>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
> >>>>>> sg_init_table(chunk->mem, HNS_ROCE_HEM_CHUNK_LEN);
> >>>>>> chunk->npages = 0;
> >>>>>> chunk->nsg = 0;
> >>>>>> + memset(chunk->vmalloc, 0, sizeof(chunk->vmalloc));
> >>>>>> list_add_tail(&chunk->list, &hem->chunk_list);
> >>>>>> }
> >>>>>>
> >>>>>> @@ -243,7 +245,15 @@ static struct hns_roce_hem
> >>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
> >>>>>> if (!buf)
> >>>>>> goto fail;
> >>>>>>
> >>>>>> - sg_set_buf(mem, buf, PAGE_SIZE << order);
> >>>>>> + if (is_vmalloc_addr(buf)) {
> >>>>>> + vmalloc = &chunk->vmalloc[chunk->npages];
> >>>>>> + vmalloc->is_vmalloc_addr = true;
> >>>>>> + vmalloc->vmalloc_addr = buf;
> >>>>>> + sg_set_page(mem, vmalloc_to_page(buf),
> >>>>>> + PAGE_SIZE << order, offset_in_page(buf));
> >>>>>> + } else {
> >>>>>> + sg_set_buf(mem, buf, PAGE_SIZE << order);
> >>>>>> + }
> >>>>>> WARN_ON(mem->offset);
> >>>>>> sg_dma_len(mem) = PAGE_SIZE << order;
> >>>>>>
> >>>>>> @@ -262,17 +272,25 @@ static struct hns_roce_hem
> >>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
> >>>>>> void hns_roce_free_hem(struct hns_roce_dev *hr_dev, struct
> >>>>>> hns_roce_hem *hem)
> >>>>>> {
> >>>>>> struct hns_roce_hem_chunk *chunk, *tmp;
> >>>>>> + void *cpu_addr;
> >>>>>> int i;
> >>>>>>
> >>>>>> if (!hem)
> >>>>>> return;
> >>>>>>
> >>>>>> list_for_each_entry_safe(chunk, tmp, &hem->chunk_list, list) {
> >>>>>> - for (i = 0; i < chunk->npages; ++i)
> >>>>>> + for (i = 0; i < chunk->npages; ++i) {
> >>>>>> + if (chunk->vmalloc[i].is_vmalloc_addr)
> >>>>>> + cpu_addr = chunk->vmalloc[i].vmalloc_addr;
> >>>>>> + else
> >>>>>> + cpu_addr =
> >>>>>> + lowmem_page_address(sg_page(&chunk->mem[i]));
> >>>>>> +
> >>>>>> dma_free_coherent(hr_dev->dev,
> >>>>>> chunk->mem[i].length,
> >>>>>> - lowmem_page_address(sg_page(&chunk->mem[i])),
> >>>>>> + cpu_addr,
> >>>>>> sg_dma_address(&chunk->mem[i]));
> >>>>>> + }
> >>>>>> kfree(chunk);
> >>>>>> }
> >>>>>>
> >>>>>> @@ -774,6 +792,12 @@ void *hns_roce_table_find(struct hns_roce_dev
> >>>>>> *hr_dev,
> >>>>>>
> >>>>>> if (chunk->mem[i].length > (u32)offset) {
> >>>>>> page = sg_page(&chunk->mem[i]);
> >>>>>> + if (chunk->vmalloc[i].is_vmalloc_addr) {
> >>>>>> + mutex_unlock(&table->mutex);
> >>>>>> + return page ?
> >>>>>> + chunk->vmalloc[i].vmalloc_addr
> >>>>>> + + offset : NULL;
> >>>>>> + }
> >>>>>> goto out;
> >>>>>> }
> >>>>>> offset -= chunk->mem[i].length;
> >>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h
> >>>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.h
> >>>>>> index af28bbf..62d712a 100644
> >>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.h
> >>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h
> >>>>>> @@ -72,11 +72,17 @@ enum {
> >>>>>> HNS_ROCE_HEM_PAGE_SIZE = 1 << HNS_ROCE_HEM_PAGE_SHIFT,
> >>>>>> };
> >>>>>>
> >>>>>> +struct hns_roce_vmalloc {
> >>>>>> + bool is_vmalloc_addr;
> >>>>>> + void *vmalloc_addr;
> >>>>>> +};
> >>>>>> +
> >>>>>> struct hns_roce_hem_chunk {
> >>>>>> struct list_head list;
> >>>>>> int npages;
> >>>>>> int nsg;
> >>>>>> struct scatterlist mem[HNS_ROCE_HEM_CHUNK_LEN];
> >>>>>> + struct hns_roce_vmalloc vmalloc[HNS_ROCE_HEM_CHUNK_LEN];
> >>>>>> };
> >>>>>>
> >>>>>> struct hns_roce_hem {
> >>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
> >>>>>> b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
> >>>>>> index b99d70a..9e19bf1 100644
> >>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
> >>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
> >>>>>> @@ -1093,9 +1093,11 @@ static int hns_roce_v2_write_mtpt(void
> >>>>>> *mb_buf, struct hns_roce_mr *mr,
> >>>>>> {
> >>>>>> struct hns_roce_v2_mpt_entry *mpt_entry;
> >>>>>> struct scatterlist *sg;
> >>>>>> + u64 page_addr = 0;
> >>>>>> u64 *pages;
> >>>>>> + int i = 0, j = 0;
> >>>>>> + int len = 0;
> >>>>>> int entry;
> >>>>>> - int i;
> >>>>>>
> >>>>>> mpt_entry = mb_buf;
> >>>>>> memset(mpt_entry, 0, sizeof(*mpt_entry));
> >>>>>> @@ -1153,14 +1155,20 @@ static int hns_roce_v2_write_mtpt(void
> >>>>>> *mb_buf, struct hns_roce_mr *mr,
> >>>>>>
> >>>>>> i = 0;
> >>>>>> for_each_sg(mr->umem->sg_head.sgl, sg, mr->umem->nmap, entry) {
> >>>>>> - pages[i] = ((u64)sg_dma_address(sg)) >> 6;
> >>>>>> -
> >>>>>> - /* Record the first 2 entry directly to MTPT table */
> >>>>>> - if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
> >>>>>> - break;
> >>>>>> - i++;
> >>>>>> + len = sg_dma_len(sg) >> PAGE_SHIFT;
> >>>>>> + for (j = 0; j < len; ++j) {
> >>>>>> + page_addr = sg_dma_address(sg) +
> >>>>>> + (j << mr->umem->page_shift);
> >>>>>> + pages[i] = page_addr >> 6;
> >>>>>> +
> >>>>>> + /* Record the first 2 entry directly to MTPT table */
> >>>>>> + if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
> >>>>>> + goto found;
> >>>>>> + i++;
> >>>>>> + }
> >>>>>> }
> >>>>>>
> >>>>>> +found:
> >>>>>> mpt_entry->pa0_l = cpu_to_le32(lower_32_bits(pages[0]));
> >>>>>> roce_set_field(mpt_entry->byte_56_pa0_h, V2_MPT_BYTE_56_PA0_H_M,
> >>>>>> V2_MPT_BYTE_56_PA0_H_S,
> >>>>>> --
> >>>>>> 1.9.1
> >>>>>>
> >>>> _______________________________________________
> >>>> iommu mailing list
> >>>> iommu@lists.linux-foundation.org
> >>>> https://lists.linuxfoundation.org/mailman/listinfo/iommu
> >>> .
> >>>
> >>
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at http://vger.kernel.org/majordomo-info.html
> >
> > .
> >
>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-11-07 2:45 ` Wei Hu (Xavier)
2017-11-07 6:32 ` Leon Romanovsky
@ 2017-11-07 15:48 ` Jason Gunthorpe
2017-11-07 15:58 ` Christoph Hellwig
2017-11-09 1:30 ` Wei Hu (Xavier)
1 sibling, 2 replies; 14+ messages in thread
From: Jason Gunthorpe @ 2017-11-07 15:48 UTC (permalink / raw)
To: Wei Hu (Xavier)
Cc: Robin Murphy, Leon Romanovsky, shaobo.xu, xavier.huwei,
lijun_nudt, oulijun, linux-rdma, charles.chenxin, linuxarm,
iommu, linux-kernel, linux-mm, dledford, liuyixian, zhangxiping3,
shaoboxu
On Tue, Nov 07, 2017 at 10:45:29AM +0800, Wei Hu (Xavier) wrote:
> We reconstruct the code as below:
> It replaces dma_alloc_coherent with __get_free_pages and
> dma_map_single functions. So, we can vmap serveral ptrs returned by
> __get_free_pages, right?
Can't you just use vmalloc and dma_map that? Other drivers follow that
approach..
However, dma_alloc_coherent and dma_map_single are not the same
thing. You can't touch the vmap memory once you call dma_map unless
the driver also includes dma cache flushing calls in all the right
places.
The difference is that alloc_coherent will return non-cachable memory
if necessary, while get_free_pages does not.
Jason
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-11-07 15:48 ` Jason Gunthorpe
@ 2017-11-07 15:58 ` Christoph Hellwig
2017-11-07 16:03 ` Jason Gunthorpe
2017-11-09 1:26 ` Wei Hu (Xavier)
2017-11-09 1:30 ` Wei Hu (Xavier)
1 sibling, 2 replies; 14+ messages in thread
From: Christoph Hellwig @ 2017-11-07 15:58 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Wei Hu (Xavier),
Robin Murphy, Leon Romanovsky, shaobo.xu, xavier.huwei,
lijun_nudt, oulijun, linux-rdma, charles.chenxin, linuxarm,
iommu, linux-kernel, linux-mm, dledford, liuyixian, zhangxiping3,
shaoboxu
On Tue, Nov 07, 2017 at 08:48:38AM -0700, Jason Gunthorpe wrote:
> Can't you just use vmalloc and dma_map that? Other drivers follow that
> approach..
You can't easily due to the flushing requirements. We used to do that
in XFS and it led to problems. You need the page allocator + vmap +
invalidate_kernel_vmap_range + flush_kernel_vmap_range to get the
cache flushing right.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-11-07 15:58 ` Christoph Hellwig
@ 2017-11-07 16:03 ` Jason Gunthorpe
2017-11-09 1:26 ` Wei Hu (Xavier)
1 sibling, 0 replies; 14+ messages in thread
From: Jason Gunthorpe @ 2017-11-07 16:03 UTC (permalink / raw)
To: Christoph Hellwig
Cc: Wei Hu (Xavier),
Robin Murphy, Leon Romanovsky, shaobo.xu, xavier.huwei,
lijun_nudt, oulijun, linux-rdma, charles.chenxin, linuxarm,
iommu, linux-kernel, linux-mm, dledford, liuyixian, zhangxiping3,
shaoboxu
On Tue, Nov 07, 2017 at 07:58:05AM -0800, Christoph Hellwig wrote:
> On Tue, Nov 07, 2017 at 08:48:38AM -0700, Jason Gunthorpe wrote:
> > Can't you just use vmalloc and dma_map that? Other drivers follow that
> > approach..
>
> You can't easily due to the flushing requirements. We used to do that
> in XFS and it led to problems. You need the page allocator + vmap +
> invalidate_kernel_vmap_range + flush_kernel_vmap_range to get the
> cache flushing right.
Yes, exactly something ugly like that.. :\
Jason
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-11-07 6:32 ` Leon Romanovsky
@ 2017-11-09 1:17 ` Wei Hu (Xavier)
0 siblings, 0 replies; 14+ messages in thread
From: Wei Hu (Xavier) @ 2017-11-09 1:17 UTC (permalink / raw)
To: Leon Romanovsky
Cc: Robin Murphy, shaobo.xu, xavier.huwei, lijun_nudt, oulijun,
linux-rdma, charles.chenxin, linuxarm, iommu, linux-kernel,
linux-mm, dledford, liuyixian, zhangxiping3, shaoboxu
On 2017/11/7 14:32, Leon Romanovsky wrote:
> On Tue, Nov 07, 2017 at 10:45:29AM +0800, Wei Hu (Xavier) wrote:
>>
>> On 2017/11/1 20:26, Robin Murphy wrote:
>>> On 01/11/17 07:46, Wei Hu (Xavier) wrote:
>>>> On 2017/10/12 20:59, Robin Murphy wrote:
>>>>> On 12/10/17 13:31, Wei Hu (Xavier) wrote:
>>>>>> On 2017/10/1 0:10, Leon Romanovsky wrote:
>>>>>>> On Sat, Sep 30, 2017 at 05:28:59PM +0800, Wei Hu (Xavier) wrote:
>>>>>>>> If the IOMMU is enabled, the length of sg obtained from
>>>>>>>> __iommu_map_sg_attrs is not 4kB. When the IOVA is set with the sg
>>>>>>>> dma address, the IOVA will not be page continuous. and the VA
>>>>>>>> returned from dma_alloc_coherent is a vmalloc address. However,
>>>>>>>> the VA obtained by the page_address is a discontinuous VA. Under
>>>>>>>> these circumstances, the IOVA should be calculated based on the
>>>>>>>> sg length, and record the VA returned from dma_alloc_coherent
>>>>>>>> in the struct of hem.
>>>>>>>>
>>>>>>>> Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
>>>>>>>> Signed-off-by: Shaobo Xu <xushaobo2@huawei.com>
>>>>>>>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>>>>>>>> ---
>>>>>>> Doug,
>>>>>>>
>>>>>>> I didn't invest time in reviewing it, but having "is_vmalloc_addr" in
>>>>>>> driver code to deal with dma_alloc_coherent is most probably wrong.
>>>>>>>
>>>>>>> Thanks
>>>>>> Hi, Leon & Doug
>>>>>> We refered the function named __ttm_dma_alloc_page in the kernel
>>>>>> code as below:
>>>>>> And there are similar methods in bch_bio_map and mem_to_page
>>>>>> functions in current 4.14-rcx.
>>>>>>
>>>>>> static struct dma_page *__ttm_dma_alloc_page(struct dma_pool *pool)
>>>>>> {
>>>>>> struct dma_page *d_page;
>>>>>>
>>>>>> d_page = kmalloc(sizeof(struct dma_page), GFP_KERNEL);
>>>>>> if (!d_page)
>>>>>> return NULL;
>>>>>>
>>>>>> d_page->vaddr = dma_alloc_coherent(pool->dev, pool->size,
>>>>>> &d_page->dma,
>>>>>> pool->gfp_flags);
>>>>>> if (d_page->vaddr) {
>>>>>> if (is_vmalloc_addr(d_page->vaddr))
>>>>>> d_page->p = vmalloc_to_page(d_page->vaddr);
>>>>>> else
>>>>>> d_page->p = virt_to_page(d_page->vaddr);
>>>>> There are cases on various architectures where neither of those is
>>>>> right. Whether those actually intersect with TTM or RDMA use-cases is
>>>>> another matter, of course.
>>>>>
>>>>> What definitely is a problem is if you ever take that page and end up
>>>>> accessing it through any virtual address other than the one explicitly
>>>>> returned by dma_alloc_coherent(). That can blow the coherency wide open
>>>>> and invite data loss, right up to killing the whole system with a
>>>>> machine check on certain architectures.
>>>>>
>>>>> Robin.
>>>> Hi, Robin
>>>> Thanks for your comment.
>>>>
>>>> We have one problem and the related code as below.
>>>> 1. call dma_alloc_coherent function serval times to alloc memory.
>>>> 2. vmap the allocated memory pages.
>>>> 3. software access memory by using the return virt addr of vmap
>>>> and hardware using the dma addr of dma_alloc_coherent.
>>> The simple answer is "don't do that". Seriously. dma_alloc_coherent()
>>> gives you a CPU virtual address and a DMA address with which to access
>>> your buffer, and that is the limit of what you may infer about it. You
>>> have no guarantee that the virtual address is either in the linear map
>>> or vmalloc, and not some other special place. You have no guarantee that
>>> the underlying memory even has an associated struct page at all.
>>>
>>>> When IOMMU is disabled in ARM64 architecture, we use virt_to_page()
>>>> before vmap(), it works. And when IOMMU is enabled using
>>>> virt_to_page() will cause calltrace later, we found the return
>>>> addr of dma_alloc_coherent is vmalloc addr, so we add the
>>>> condition judgement statement as below, it works.
>>>> for (i = 0; i < buf->nbufs; ++i)
>>>> pages[i] =
>>>> is_vmalloc_addr(buf->page_list[i].buf) ?
>>>> vmalloc_to_page(buf->page_list[i].buf) :
>>>> virt_to_page(buf->page_list[i].buf);
>>>> Can you give us suggestion? better method?
>>> Oh my goodness, having now taken a closer look at this driver, I'm lost
>>> for words in disbelief. To pick just one example:
>>>
>>> u32 bits_per_long = BITS_PER_LONG;
>>> ...
>>> if (bits_per_long == 64) {
>>> /* memory mapping nonsense */
>>> }
>>>
>>> WTF does the size of a long have to do with DMA buffer management!?
>>>
>>> Of course I can guess that it might be trying to make some tortuous
>>> inference about vmalloc space being constrained on 32-bit platforms, but
>>> still...
>>>
>>>> The related code as below:
>>>> buf->page_list = kcalloc(buf->nbufs, sizeof(*buf->page_list),
>>>> GFP_KERNEL);
>>>> if (!buf->page_list)
>>>> return -ENOMEM;
>>>>
>>>> for (i = 0; i < buf->nbufs; ++i) {
>>>> buf->page_list[i].buf = dma_alloc_coherent(dev,
>>>> page_size, &t,
>>>> GFP_KERNEL);
>>>> if (!buf->page_list[i].buf)
>>>> goto err_free;
>>>>
>>>> buf->page_list[i].map = t;
>>>> memset(buf->page_list[i].buf, 0, page_size);
>>>> }
>>>>
>>>> pages = kmalloc_array(buf->nbufs, sizeof(*pages),
>>>> GFP_KERNEL);
>>>> if (!pages)
>>>> goto err_free;
>>>>
>>>> for (i = 0; i < buf->nbufs; ++i)
>>>> pages[i] =
>>>> is_vmalloc_addr(buf->page_list[i].buf) ?
>>>> vmalloc_to_page(buf->page_list[i].buf) :
>>>> virt_to_page(buf->page_list[i].buf);
>>>>
>>>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>>>> PAGE_KERNEL);
>>>> kfree(pages);
>>>> if (!buf->direct.buf)
>>>> goto err_free;
>>> OK, this is complete crap. As above, you cannot assume that a struct
>>> page even exists; even if it does you cannot assume that using a
>>> PAGE_KERNEL mapping will not result in mismatched attributes,
>>> unpredictable behaviour and data loss. Trying to remap coherent DMA
>>> allocations like this is just egregiously wrong.
>>>
>>> What I do like is that you can seemingly fix all this by simply deleting
>>> hns_roce_buf::direct and all the garbage code related to it, and using
>>> the page_list entries consistently because the alternate paths involving
>>> those appear to do the right thing already.
>>>
>>> That is, of course, assuming that the buffers involved can be so large
>>> that it's not practical to just always make a single allocation and
>>> fragment it into multiple descriptors if the hardware does have some
>>> maximum length constraint - frankly I'm a little puzzled by the
>>> PAGE_SIZE * 2 threshold, given that that's not a fixed size.
>>>
>>> Robin.
>> Hii 1/4 ?Robin
>>
>> We reconstruct the code as below:
>> It replaces dma_alloc_coherent with __get_free_pages and
>> dma_map_single
>> functions. So, we can vmap serveral ptrs returned by
>> __get_free_pages, right?
> Most probably not, you should get rid of your virt_to_page/vmap calls.
>
> Thanks
Hi, Leon
Thanks for your suggestion.
I will send a patch to fix it.
Regards
Wei Hu
>>
>> buf->page_list = kcalloc(buf->nbufs, sizeof(*buf->page_list),
>> GFP_KERNEL);
>> if (!buf->page_list)
>> return -ENOMEM;
>>
>> for (i = 0; i < buf->nbufs; ++i) {
>> ptr = (void *)__get_free_pages(GFP_KERNEL | __GFP_ZERO,
>> get_order(page_size));
>> if (!ptr) {
>> dev_err(dev, "Alloc pages error.\n");
>> goto err_free;
>> }
>>
>> t = dma_map_single(dev, ptr, page_size,
>> DMA_BIDIRECTIONAL);
>> if (dma_mapping_error(dev, t)) {
>> dev_err(dev, "DMA mapping error.\n");
>> free_pages((unsigned long)ptr,
>> get_order(page_size));
>> goto err_free;
>> }
>>
>> buf->page_list[i].buf = ptr;
>> buf->page_list[i].map = t;
>> }
>>
>> pages = kmalloc_array(buf->nbufs, sizeof(*pages),
>> GFP_KERNEL);
>> if (!pages)
>> goto err_free;
>>
>> for (i = 0; i < buf->nbufs; ++i)
>> pages[i] = virt_to_page(buf->page_list[i].buf);
>>
>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>> PAGE_KERNEL);
>> kfree(pages);
>> if (!buf->direct.buf)
>> goto err_free;
>>
>>
>> Regards
>> Wei Hu
>>>> Regards
>>>> Wei Hu
>>>>>> } else {
>>>>>> kfree(d_page);
>>>>>> d_page = NULL;
>>>>>> }
>>>>>> return d_page;
>>>>>> }
>>>>>>
>>>>>> Regards
>>>>>> Wei Hu
>>>>>>>> drivers/infiniband/hw/hns/hns_roce_alloc.c | 5 ++++-
>>>>>>>> drivers/infiniband/hw/hns/hns_roce_hem.c | 30
>>>>>>>> +++++++++++++++++++++++++++---
>>>>>>>> drivers/infiniband/hw/hns/hns_roce_hem.h | 6 ++++++
>>>>>>>> drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 22 +++++++++++++++-------
>>>>>>>> 4 files changed, 52 insertions(+), 11 deletions(-)
>>>>>>>>
>>>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>>>> b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>>>> index 3e4c525..a69cd4b 100644
>>>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>>>> @@ -243,7 +243,10 @@ int hns_roce_buf_alloc(struct hns_roce_dev
>>>>>>>> *hr_dev, u32 size, u32 max_direct,
>>>>>>>> goto err_free;
>>>>>>>>
>>>>>>>> for (i = 0; i < buf->nbufs; ++i)
>>>>>>>> - pages[i] = virt_to_page(buf->page_list[i].buf);
>>>>>>>> + pages[i] =
>>>>>>>> + is_vmalloc_addr(buf->page_list[i].buf) ?
>>>>>>>> + vmalloc_to_page(buf->page_list[i].buf) :
>>>>>>>> + virt_to_page(buf->page_list[i].buf);
>>>>>>>>
>>>>>>>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>>>>>>>> PAGE_KERNEL);
>>>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>>>> index 8388ae2..4a3d1d4 100644
>>>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>>>> @@ -200,6 +200,7 @@ static struct hns_roce_hem
>>>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>>>> gfp_t gfp_mask)
>>>>>>>> {
>>>>>>>> struct hns_roce_hem_chunk *chunk = NULL;
>>>>>>>> + struct hns_roce_vmalloc *vmalloc;
>>>>>>>> struct hns_roce_hem *hem;
>>>>>>>> struct scatterlist *mem;
>>>>>>>> int order;
>>>>>>>> @@ -227,6 +228,7 @@ static struct hns_roce_hem
>>>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>>>> sg_init_table(chunk->mem, HNS_ROCE_HEM_CHUNK_LEN);
>>>>>>>> chunk->npages = 0;
>>>>>>>> chunk->nsg = 0;
>>>>>>>> + memset(chunk->vmalloc, 0, sizeof(chunk->vmalloc));
>>>>>>>> list_add_tail(&chunk->list, &hem->chunk_list);
>>>>>>>> }
>>>>>>>>
>>>>>>>> @@ -243,7 +245,15 @@ static struct hns_roce_hem
>>>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>>>> if (!buf)
>>>>>>>> goto fail;
>>>>>>>>
>>>>>>>> - sg_set_buf(mem, buf, PAGE_SIZE << order);
>>>>>>>> + if (is_vmalloc_addr(buf)) {
>>>>>>>> + vmalloc = &chunk->vmalloc[chunk->npages];
>>>>>>>> + vmalloc->is_vmalloc_addr = true;
>>>>>>>> + vmalloc->vmalloc_addr = buf;
>>>>>>>> + sg_set_page(mem, vmalloc_to_page(buf),
>>>>>>>> + PAGE_SIZE << order, offset_in_page(buf));
>>>>>>>> + } else {
>>>>>>>> + sg_set_buf(mem, buf, PAGE_SIZE << order);
>>>>>>>> + }
>>>>>>>> WARN_ON(mem->offset);
>>>>>>>> sg_dma_len(mem) = PAGE_SIZE << order;
>>>>>>>>
>>>>>>>> @@ -262,17 +272,25 @@ static struct hns_roce_hem
>>>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>>>> void hns_roce_free_hem(struct hns_roce_dev *hr_dev, struct
>>>>>>>> hns_roce_hem *hem)
>>>>>>>> {
>>>>>>>> struct hns_roce_hem_chunk *chunk, *tmp;
>>>>>>>> + void *cpu_addr;
>>>>>>>> int i;
>>>>>>>>
>>>>>>>> if (!hem)
>>>>>>>> return;
>>>>>>>>
>>>>>>>> list_for_each_entry_safe(chunk, tmp, &hem->chunk_list, list) {
>>>>>>>> - for (i = 0; i < chunk->npages; ++i)
>>>>>>>> + for (i = 0; i < chunk->npages; ++i) {
>>>>>>>> + if (chunk->vmalloc[i].is_vmalloc_addr)
>>>>>>>> + cpu_addr = chunk->vmalloc[i].vmalloc_addr;
>>>>>>>> + else
>>>>>>>> + cpu_addr =
>>>>>>>> + lowmem_page_address(sg_page(&chunk->mem[i]));
>>>>>>>> +
>>>>>>>> dma_free_coherent(hr_dev->dev,
>>>>>>>> chunk->mem[i].length,
>>>>>>>> - lowmem_page_address(sg_page(&chunk->mem[i])),
>>>>>>>> + cpu_addr,
>>>>>>>> sg_dma_address(&chunk->mem[i]));
>>>>>>>> + }
>>>>>>>> kfree(chunk);
>>>>>>>> }
>>>>>>>>
>>>>>>>> @@ -774,6 +792,12 @@ void *hns_roce_table_find(struct hns_roce_dev
>>>>>>>> *hr_dev,
>>>>>>>>
>>>>>>>> if (chunk->mem[i].length > (u32)offset) {
>>>>>>>> page = sg_page(&chunk->mem[i]);
>>>>>>>> + if (chunk->vmalloc[i].is_vmalloc_addr) {
>>>>>>>> + mutex_unlock(&table->mutex);
>>>>>>>> + return page ?
>>>>>>>> + chunk->vmalloc[i].vmalloc_addr
>>>>>>>> + + offset : NULL;
>>>>>>>> + }
>>>>>>>> goto out;
>>>>>>>> }
>>>>>>>> offset -= chunk->mem[i].length;
>>>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>>>> index af28bbf..62d712a 100644
>>>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>>>> @@ -72,11 +72,17 @@ enum {
>>>>>>>> HNS_ROCE_HEM_PAGE_SIZE = 1 << HNS_ROCE_HEM_PAGE_SHIFT,
>>>>>>>> };
>>>>>>>>
>>>>>>>> +struct hns_roce_vmalloc {
>>>>>>>> + bool is_vmalloc_addr;
>>>>>>>> + void *vmalloc_addr;
>>>>>>>> +};
>>>>>>>> +
>>>>>>>> struct hns_roce_hem_chunk {
>>>>>>>> struct list_head list;
>>>>>>>> int npages;
>>>>>>>> int nsg;
>>>>>>>> struct scatterlist mem[HNS_ROCE_HEM_CHUNK_LEN];
>>>>>>>> + struct hns_roce_vmalloc vmalloc[HNS_ROCE_HEM_CHUNK_LEN];
>>>>>>>> };
>>>>>>>>
>>>>>>>> struct hns_roce_hem {
>>>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>>>> b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>>>> index b99d70a..9e19bf1 100644
>>>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>>>> @@ -1093,9 +1093,11 @@ static int hns_roce_v2_write_mtpt(void
>>>>>>>> *mb_buf, struct hns_roce_mr *mr,
>>>>>>>> {
>>>>>>>> struct hns_roce_v2_mpt_entry *mpt_entry;
>>>>>>>> struct scatterlist *sg;
>>>>>>>> + u64 page_addr = 0;
>>>>>>>> u64 *pages;
>>>>>>>> + int i = 0, j = 0;
>>>>>>>> + int len = 0;
>>>>>>>> int entry;
>>>>>>>> - int i;
>>>>>>>>
>>>>>>>> mpt_entry = mb_buf;
>>>>>>>> memset(mpt_entry, 0, sizeof(*mpt_entry));
>>>>>>>> @@ -1153,14 +1155,20 @@ static int hns_roce_v2_write_mtpt(void
>>>>>>>> *mb_buf, struct hns_roce_mr *mr,
>>>>>>>>
>>>>>>>> i = 0;
>>>>>>>> for_each_sg(mr->umem->sg_head.sgl, sg, mr->umem->nmap, entry) {
>>>>>>>> - pages[i] = ((u64)sg_dma_address(sg)) >> 6;
>>>>>>>> -
>>>>>>>> - /* Record the first 2 entry directly to MTPT table */
>>>>>>>> - if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>>>>>>> - break;
>>>>>>>> - i++;
>>>>>>>> + len = sg_dma_len(sg) >> PAGE_SHIFT;
>>>>>>>> + for (j = 0; j < len; ++j) {
>>>>>>>> + page_addr = sg_dma_address(sg) +
>>>>>>>> + (j << mr->umem->page_shift);
>>>>>>>> + pages[i] = page_addr >> 6;
>>>>>>>> +
>>>>>>>> + /* Record the first 2 entry directly to MTPT table */
>>>>>>>> + if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>>>>>>> + goto found;
>>>>>>>> + i++;
>>>>>>>> + }
>>>>>>>> }
>>>>>>>>
>>>>>>>> +found:
>>>>>>>> mpt_entry->pa0_l = cpu_to_le32(lower_32_bits(pages[0]));
>>>>>>>> roce_set_field(mpt_entry->byte_56_pa0_h, V2_MPT_BYTE_56_PA0_H_M,
>>>>>>>> V2_MPT_BYTE_56_PA0_H_S,
>>>>>>>> --
>>>>>>>> 1.9.1
>>>>>>>>
>>>>>> _______________________________________________
>>>>>> iommu mailing list
>>>>>> iommu@lists.linux-foundation.org
>>>>>> https://lists.linuxfoundation.org/mailman/listinfo/iommu
>>>>> .
>>>>>
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
>>> the body of a message to majordomo@vger.kernel.org
>>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>>>
>>> .
>>>
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-11-07 15:58 ` Christoph Hellwig
2017-11-07 16:03 ` Jason Gunthorpe
@ 2017-11-09 1:26 ` Wei Hu (Xavier)
1 sibling, 0 replies; 14+ messages in thread
From: Wei Hu (Xavier) @ 2017-11-09 1:26 UTC (permalink / raw)
To: Christoph Hellwig, Jason Gunthorpe
Cc: Robin Murphy, Leon Romanovsky, shaobo.xu, xavier.huwei,
lijun_nudt, oulijun, linux-rdma, charles.chenxin, linuxarm,
iommu, linux-kernel, linux-mm, dledford, liuyixian, zhangxiping3,
shaoboxu
On 2017/11/7 23:58, Christoph Hellwig wrote:
> On Tue, Nov 07, 2017 at 08:48:38AM -0700, Jason Gunthorpe wrote:
>> Can't you just use vmalloc and dma_map that? Other drivers follow that
>> approach..
> You can't easily due to the flushing requirements. We used to do that
> in XFS and it led to problems. You need the page allocator + vmap +
> invalidate_kernel_vmap_range + flush_kernel_vmap_range to get the
> cache flushing right.
>
> .
Hi, Christoph Hellwig
Thanks for your suggestion.
Regards
Wei Hu
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-11-07 15:48 ` Jason Gunthorpe
2017-11-07 15:58 ` Christoph Hellwig
@ 2017-11-09 1:30 ` Wei Hu (Xavier)
1 sibling, 0 replies; 14+ messages in thread
From: Wei Hu (Xavier) @ 2017-11-09 1:30 UTC (permalink / raw)
To: Jason Gunthorpe
Cc: Robin Murphy, Leon Romanovsky, shaobo.xu, xavier.huwei,
lijun_nudt, oulijun, linux-rdma, charles.chenxin, linuxarm,
iommu, linux-kernel, linux-mm, dledford, liuyixian, zhangxiping3,
shaoboxu
On 2017/11/7 23:48, Jason Gunthorpe wrote:
> On Tue, Nov 07, 2017 at 10:45:29AM +0800, Wei Hu (Xavier) wrote:
>
>> We reconstruct the code as below:
>> It replaces dma_alloc_coherent with __get_free_pages and
>> dma_map_single functions. So, we can vmap serveral ptrs returned by
>> __get_free_pages, right?
> Can't you just use vmalloc and dma_map that? Other drivers follow that
> approach..
>
> However, dma_alloc_coherent and dma_map_single are not the same
> thing. You can't touch the vmap memory once you call dma_map unless
> the driver also includes dma cache flushing calls in all the right
> places.
>
> The difference is that alloc_coherent will return non-cachable memory
> if necessary, while get_free_pages does not.
>
> Jason
Hi, Jason
Thanks for your suggestion.
We will fix it.
Regards
Wei Hu
>
> .
>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
* Re: [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08
2017-11-01 12:26 ` Robin Murphy
2017-11-07 2:45 ` Wei Hu (Xavier)
@ 2017-11-09 1:36 ` Wei Hu (Xavier)
1 sibling, 0 replies; 14+ messages in thread
From: Wei Hu (Xavier) @ 2017-11-09 1:36 UTC (permalink / raw)
To: Robin Murphy
Cc: Leon Romanovsky, shaobo.xu, xavier.huwei, lijun_nudt, oulijun,
linux-rdma, charles.chenxin, linuxarm, iommu, linux-kernel,
linux-mm, dledford, liuyixian, zhangxiping3, shaoboxu
On 2017/11/1 20:26, Robin Murphy wrote:
> On 01/11/17 07:46, Wei Hu (Xavier) wrote:
>>
>> On 2017/10/12 20:59, Robin Murphy wrote:
>>> On 12/10/17 13:31, Wei Hu (Xavier) wrote:
>>>> On 2017/10/1 0:10, Leon Romanovsky wrote:
>>>>> On Sat, Sep 30, 2017 at 05:28:59PM +0800, Wei Hu (Xavier) wrote:
>>>>>> If the IOMMU is enabled, the length of sg obtained from
>>>>>> __iommu_map_sg_attrs is not 4kB. When the IOVA is set with the sg
>>>>>> dma address, the IOVA will not be page continuous. and the VA
>>>>>> returned from dma_alloc_coherent is a vmalloc address. However,
>>>>>> the VA obtained by the page_address is a discontinuous VA. Under
>>>>>> these circumstances, the IOVA should be calculated based on the
>>>>>> sg length, and record the VA returned from dma_alloc_coherent
>>>>>> in the struct of hem.
>>>>>>
>>>>>> Signed-off-by: Wei Hu (Xavier) <xavier.huwei@huawei.com>
>>>>>> Signed-off-by: Shaobo Xu <xushaobo2@huawei.com>
>>>>>> Signed-off-by: Lijun Ou <oulijun@huawei.com>
>>>>>> ---
>>>>> Doug,
>>>>>
>>>>> I didn't invest time in reviewing it, but having "is_vmalloc_addr" in
>>>>> driver code to deal with dma_alloc_coherent is most probably wrong.
>>>>>
>>>>> Thanks
>>>> Hi, Leon & Doug
>>>> We refered the function named __ttm_dma_alloc_page in the kernel
>>>> code as below:
>>>> And there are similar methods in bch_bio_map and mem_to_page
>>>> functions in current 4.14-rcx.
>>>>
>>>> static struct dma_page *__ttm_dma_alloc_page(struct dma_pool *pool)
>>>> {
>>>> struct dma_page *d_page;
>>>>
>>>> d_page = kmalloc(sizeof(struct dma_page), GFP_KERNEL);
>>>> if (!d_page)
>>>> return NULL;
>>>>
>>>> d_page->vaddr = dma_alloc_coherent(pool->dev, pool->size,
>>>> &d_page->dma,
>>>> pool->gfp_flags);
>>>> if (d_page->vaddr) {
>>>> if (is_vmalloc_addr(d_page->vaddr))
>>>> d_page->p = vmalloc_to_page(d_page->vaddr);
>>>> else
>>>> d_page->p = virt_to_page(d_page->vaddr);
>>> There are cases on various architectures where neither of those is
>>> right. Whether those actually intersect with TTM or RDMA use-cases is
>>> another matter, of course.
>>>
>>> What definitely is a problem is if you ever take that page and end up
>>> accessing it through any virtual address other than the one explicitly
>>> returned by dma_alloc_coherent(). That can blow the coherency wide open
>>> and invite data loss, right up to killing the whole system with a
>>> machine check on certain architectures.
>>>
>>> Robin.
>> Hi, Robin
>> Thanks for your comment.
>>
>> We have one problem and the related code as below.
>> 1. call dma_alloc_coherent function serval times to alloc memory.
>> 2. vmap the allocated memory pages.
>> 3. software access memory by using the return virt addr of vmap
>> and hardware using the dma addr of dma_alloc_coherent.
> The simple answer is "don't do that". Seriously. dma_alloc_coherent()
> gives you a CPU virtual address and a DMA address with which to access
> your buffer, and that is the limit of what you may infer about it. You
> have no guarantee that the virtual address is either in the linear map
> or vmalloc, and not some other special place. You have no guarantee that
> the underlying memory even has an associated struct page at all.
>
>> When IOMMU is disabled in ARM64 architecture, we use virt_to_page()
>> before vmap(), it works. And when IOMMU is enabled using
>> virt_to_page() will cause calltrace later, we found the return
>> addr of dma_alloc_coherent is vmalloc addr, so we add the
>> condition judgement statement as below, it works.
>> for (i = 0; i < buf->nbufs; ++i)
>> pages[i] =
>> is_vmalloc_addr(buf->page_list[i].buf) ?
>> vmalloc_to_page(buf->page_list[i].buf) :
>> virt_to_page(buf->page_list[i].buf);
>> Can you give us suggestion? better method?
> Oh my goodness, having now taken a closer look at this driver, I'm lost
> for words in disbelief. To pick just one example:
>
> u32 bits_per_long = BITS_PER_LONG;
> ...
> if (bits_per_long == 64) {
> /* memory mapping nonsense */
> }
>
> WTF does the size of a long have to do with DMA buffer management!?
>
> Of course I can guess that it might be trying to make some tortuous
> inference about vmalloc space being constrained on 32-bit platforms, but
> still...
We will fix it. Thanks
>> The related code as below:
>> buf->page_list = kcalloc(buf->nbufs, sizeof(*buf->page_list),
>> GFP_KERNEL);
>> if (!buf->page_list)
>> return -ENOMEM;
>>
>> for (i = 0; i < buf->nbufs; ++i) {
>> buf->page_list[i].buf = dma_alloc_coherent(dev,
>> page_size, &t,
>> GFP_KERNEL);
>> if (!buf->page_list[i].buf)
>> goto err_free;
>>
>> buf->page_list[i].map = t;
>> memset(buf->page_list[i].buf, 0, page_size);
>> }
>>
>> pages = kmalloc_array(buf->nbufs, sizeof(*pages),
>> GFP_KERNEL);
>> if (!pages)
>> goto err_free;
>>
>> for (i = 0; i < buf->nbufs; ++i)
>> pages[i] =
>> is_vmalloc_addr(buf->page_list[i].buf) ?
>> vmalloc_to_page(buf->page_list[i].buf) :
>> virt_to_page(buf->page_list[i].buf);
>>
>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>> PAGE_KERNEL);
>> kfree(pages);
>> if (!buf->direct.buf)
>> goto err_free;
> OK, this is complete crap. As above, you cannot assume that a struct
> page even exists; even if it does you cannot assume that using a
> PAGE_KERNEL mapping will not result in mismatched attributes,
> unpredictable behaviour and data loss. Trying to remap coherent DMA
> allocations like this is just egregiously wrong.
>
> What I do like is that you can seemingly fix all this by simply deleting
> hns_roce_buf::direct and all the garbage code related to it, and using
> the page_list entries consistently because the alternate paths involving
> those appear to do the right thing already.
Hi, Robin
Thanks for your suggestion.
We will fix it.
Regards
Wei Hu
> That is, of course, assuming that the buffers involved can be so large
> that it's not practical to just always make a single allocation and
> fragment it into multiple descriptors if the hardware does have some
> maximum length constraint - frankly I'm a little puzzled by the
> PAGE_SIZE * 2 threshold, given that that's not a fixed size.
>
> Robin.
>
>> Regards
>> Wei Hu
>>>> } else {
>>>> kfree(d_page);
>>>> d_page = NULL;
>>>> }
>>>> return d_page;
>>>> }
>>>>
>>>> Regards
>>>> Wei Hu
>>>>>> drivers/infiniband/hw/hns/hns_roce_alloc.c | 5 ++++-
>>>>>> drivers/infiniband/hw/hns/hns_roce_hem.c | 30
>>>>>> +++++++++++++++++++++++++++---
>>>>>> drivers/infiniband/hw/hns/hns_roce_hem.h | 6 ++++++
>>>>>> drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 22 +++++++++++++++-------
>>>>>> 4 files changed, 52 insertions(+), 11 deletions(-)
>>>>>>
>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>> b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>> index 3e4c525..a69cd4b 100644
>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c
>>>>>> @@ -243,7 +243,10 @@ int hns_roce_buf_alloc(struct hns_roce_dev
>>>>>> *hr_dev, u32 size, u32 max_direct,
>>>>>> goto err_free;
>>>>>>
>>>>>> for (i = 0; i < buf->nbufs; ++i)
>>>>>> - pages[i] = virt_to_page(buf->page_list[i].buf);
>>>>>> + pages[i] =
>>>>>> + is_vmalloc_addr(buf->page_list[i].buf) ?
>>>>>> + vmalloc_to_page(buf->page_list[i].buf) :
>>>>>> + virt_to_page(buf->page_list[i].buf);
>>>>>>
>>>>>> buf->direct.buf = vmap(pages, buf->nbufs, VM_MAP,
>>>>>> PAGE_KERNEL);
>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>> index 8388ae2..4a3d1d4 100644
>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c
>>>>>> @@ -200,6 +200,7 @@ static struct hns_roce_hem
>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>> gfp_t gfp_mask)
>>>>>> {
>>>>>> struct hns_roce_hem_chunk *chunk = NULL;
>>>>>> + struct hns_roce_vmalloc *vmalloc;
>>>>>> struct hns_roce_hem *hem;
>>>>>> struct scatterlist *mem;
>>>>>> int order;
>>>>>> @@ -227,6 +228,7 @@ static struct hns_roce_hem
>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>> sg_init_table(chunk->mem, HNS_ROCE_HEM_CHUNK_LEN);
>>>>>> chunk->npages = 0;
>>>>>> chunk->nsg = 0;
>>>>>> + memset(chunk->vmalloc, 0, sizeof(chunk->vmalloc));
>>>>>> list_add_tail(&chunk->list, &hem->chunk_list);
>>>>>> }
>>>>>>
>>>>>> @@ -243,7 +245,15 @@ static struct hns_roce_hem
>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>> if (!buf)
>>>>>> goto fail;
>>>>>>
>>>>>> - sg_set_buf(mem, buf, PAGE_SIZE << order);
>>>>>> + if (is_vmalloc_addr(buf)) {
>>>>>> + vmalloc = &chunk->vmalloc[chunk->npages];
>>>>>> + vmalloc->is_vmalloc_addr = true;
>>>>>> + vmalloc->vmalloc_addr = buf;
>>>>>> + sg_set_page(mem, vmalloc_to_page(buf),
>>>>>> + PAGE_SIZE << order, offset_in_page(buf));
>>>>>> + } else {
>>>>>> + sg_set_buf(mem, buf, PAGE_SIZE << order);
>>>>>> + }
>>>>>> WARN_ON(mem->offset);
>>>>>> sg_dma_len(mem) = PAGE_SIZE << order;
>>>>>>
>>>>>> @@ -262,17 +272,25 @@ static struct hns_roce_hem
>>>>>> *hns_roce_alloc_hem(struct hns_roce_dev *hr_dev,
>>>>>> void hns_roce_free_hem(struct hns_roce_dev *hr_dev, struct
>>>>>> hns_roce_hem *hem)
>>>>>> {
>>>>>> struct hns_roce_hem_chunk *chunk, *tmp;
>>>>>> + void *cpu_addr;
>>>>>> int i;
>>>>>>
>>>>>> if (!hem)
>>>>>> return;
>>>>>>
>>>>>> list_for_each_entry_safe(chunk, tmp, &hem->chunk_list, list) {
>>>>>> - for (i = 0; i < chunk->npages; ++i)
>>>>>> + for (i = 0; i < chunk->npages; ++i) {
>>>>>> + if (chunk->vmalloc[i].is_vmalloc_addr)
>>>>>> + cpu_addr = chunk->vmalloc[i].vmalloc_addr;
>>>>>> + else
>>>>>> + cpu_addr =
>>>>>> + lowmem_page_address(sg_page(&chunk->mem[i]));
>>>>>> +
>>>>>> dma_free_coherent(hr_dev->dev,
>>>>>> chunk->mem[i].length,
>>>>>> - lowmem_page_address(sg_page(&chunk->mem[i])),
>>>>>> + cpu_addr,
>>>>>> sg_dma_address(&chunk->mem[i]));
>>>>>> + }
>>>>>> kfree(chunk);
>>>>>> }
>>>>>>
>>>>>> @@ -774,6 +792,12 @@ void *hns_roce_table_find(struct hns_roce_dev
>>>>>> *hr_dev,
>>>>>>
>>>>>> if (chunk->mem[i].length > (u32)offset) {
>>>>>> page = sg_page(&chunk->mem[i]);
>>>>>> + if (chunk->vmalloc[i].is_vmalloc_addr) {
>>>>>> + mutex_unlock(&table->mutex);
>>>>>> + return page ?
>>>>>> + chunk->vmalloc[i].vmalloc_addr
>>>>>> + + offset : NULL;
>>>>>> + }
>>>>>> goto out;
>>>>>> }
>>>>>> offset -= chunk->mem[i].length;
>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>> b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>> index af28bbf..62d712a 100644
>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h
>>>>>> @@ -72,11 +72,17 @@ enum {
>>>>>> HNS_ROCE_HEM_PAGE_SIZE = 1 << HNS_ROCE_HEM_PAGE_SHIFT,
>>>>>> };
>>>>>>
>>>>>> +struct hns_roce_vmalloc {
>>>>>> + bool is_vmalloc_addr;
>>>>>> + void *vmalloc_addr;
>>>>>> +};
>>>>>> +
>>>>>> struct hns_roce_hem_chunk {
>>>>>> struct list_head list;
>>>>>> int npages;
>>>>>> int nsg;
>>>>>> struct scatterlist mem[HNS_ROCE_HEM_CHUNK_LEN];
>>>>>> + struct hns_roce_vmalloc vmalloc[HNS_ROCE_HEM_CHUNK_LEN];
>>>>>> };
>>>>>>
>>>>>> struct hns_roce_hem {
>>>>>> diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>> b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>> index b99d70a..9e19bf1 100644
>>>>>> --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>> +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
>>>>>> @@ -1093,9 +1093,11 @@ static int hns_roce_v2_write_mtpt(void
>>>>>> *mb_buf, struct hns_roce_mr *mr,
>>>>>> {
>>>>>> struct hns_roce_v2_mpt_entry *mpt_entry;
>>>>>> struct scatterlist *sg;
>>>>>> + u64 page_addr = 0;
>>>>>> u64 *pages;
>>>>>> + int i = 0, j = 0;
>>>>>> + int len = 0;
>>>>>> int entry;
>>>>>> - int i;
>>>>>>
>>>>>> mpt_entry = mb_buf;
>>>>>> memset(mpt_entry, 0, sizeof(*mpt_entry));
>>>>>> @@ -1153,14 +1155,20 @@ static int hns_roce_v2_write_mtpt(void
>>>>>> *mb_buf, struct hns_roce_mr *mr,
>>>>>>
>>>>>> i = 0;
>>>>>> for_each_sg(mr->umem->sg_head.sgl, sg, mr->umem->nmap, entry) {
>>>>>> - pages[i] = ((u64)sg_dma_address(sg)) >> 6;
>>>>>> -
>>>>>> - /* Record the first 2 entry directly to MTPT table */
>>>>>> - if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>>>>> - break;
>>>>>> - i++;
>>>>>> + len = sg_dma_len(sg) >> PAGE_SHIFT;
>>>>>> + for (j = 0; j < len; ++j) {
>>>>>> + page_addr = sg_dma_address(sg) +
>>>>>> + (j << mr->umem->page_shift);
>>>>>> + pages[i] = page_addr >> 6;
>>>>>> +
>>>>>> + /* Record the first 2 entry directly to MTPT table */
>>>>>> + if (i >= HNS_ROCE_V2_MAX_INNER_MTPT_NUM - 1)
>>>>>> + goto found;
>>>>>> + i++;
>>>>>> + }
>>>>>> }
>>>>>>
>>>>>> +found:
>>>>>> mpt_entry->pa0_l = cpu_to_le32(lower_32_bits(pages[0]));
>>>>>> roce_set_field(mpt_entry->byte_56_pa0_h, V2_MPT_BYTE_56_PA0_H_M,
>>>>>> V2_MPT_BYTE_56_PA0_H_S,
>>>>>> --
>>>>>> 1.9.1
>>>>>>
>>>> _______________________________________________
>>>> iommu mailing list
>>>> iommu@lists.linux-foundation.org
>>>> https://lists.linuxfoundation.org/mailman/listinfo/iommu
>>> .
>>>
>>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at http://vger.kernel.org/majordomo-info.html
>
> .
>
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 14+ messages in thread
end of thread, other threads:[~2017-11-09 1:36 UTC | newest]
Thread overview: 14+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
[not found] <1506763741-81429-1-git-send-email-xavier.huwei@huawei.com>
[not found] ` <1506763741-81429-3-git-send-email-xavier.huwei@huawei.com>
[not found] ` <20170930161023.GI2965@mtr-leonro.local>
2017-10-12 12:31 ` [PATCH for-next 2/4] RDMA/hns: Add IOMMU enable support in hip08 Wei Hu (Xavier)
2017-10-12 12:59 ` Robin Murphy
2017-11-01 7:46 ` Wei Hu (Xavier)
2017-11-01 12:26 ` Robin Murphy
2017-11-07 2:45 ` Wei Hu (Xavier)
2017-11-07 6:32 ` Leon Romanovsky
2017-11-09 1:17 ` Wei Hu (Xavier)
2017-11-07 15:48 ` Jason Gunthorpe
2017-11-07 15:58 ` Christoph Hellwig
2017-11-07 16:03 ` Jason Gunthorpe
2017-11-09 1:26 ` Wei Hu (Xavier)
2017-11-09 1:30 ` Wei Hu (Xavier)
2017-11-09 1:36 ` Wei Hu (Xavier)
2017-10-12 14:54 ` Leon Romanovsky
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).