xiaomi-sm8450/android_kernel_xiaomi_sm8450
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

IB: Refactor umem to use linear SG table

Browse Source 
This patch refactors the IB core umem code and vendor drivers to use a
linear (chained) SG table instead of chunk list.  With this change the
relevant code becomes clearer—no need for nested loops to build and
use umem.

Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This commit is contained in:
Yishai Hadas
2014-01-28 13:40:15 +02:00
committed by Roland Dreier
parent cfbf8d4857
commit eeb8461e36
16 changed files with 446 additions and 550 deletions
Download Patch File Download Diff File Expand all files Collapse all files

118
drivers/infiniband/core/umem.c
View File

@@ -42,29 +42,29 @@
#include "uverbs.h"
#define IB_UMEM_MAX_PAGE_CHUNK \
((PAGE_SIZE - offsetof(struct ib_umem_chunk, page_list)) / \
((void *) &((struct ib_umem_chunk *) 0)->page_list[1] - \
(void *) &((struct ib_umem_chunk *) 0)->page_list[0]))
static void __ib_umem_release(struct ib_device *dev, struct ib_umem *umem, int dirty)
{
struct ib_umem_chunk *chunk, *tmp;
struct scatterlist *sg;
struct page *page;
int i;
list_for_each_entry_safe(chunk, tmp, &umem->chunk_list, list) {
ib_dma_unmap_sg(dev, chunk->page_list,
chunk->nents, DMA_BIDIRECTIONAL);
for (i = 0; i < chunk->nents; ++i) {
struct page *page = sg_page(&chunk->page_list[i]);
if (umem->nmap > 0)
ib_dma_unmap_sg(dev, umem->sg_head.sgl,
umem->nmap,
DMA_BIDIRECTIONAL);
if (umem->writable && dirty)
set_page_dirty_lock(page);
put_page(page);
}
for_each_sg(umem->sg_head.sgl, sg, umem->npages, i) {
kfree(chunk);
page = sg_page(sg);
if (umem->writable && dirty)
set_page_dirty_lock(page);
put_page(page);
}
sg_free_table(&umem->sg_head);
return;
}
/**
@@ -81,15 +81,15 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
struct ib_umem *umem;
struct page **page_list;
struct vm_area_struct **vma_list;
struct ib_umem_chunk *chunk;
unsigned long locked;
unsigned long lock_limit;
unsigned long cur_base;
unsigned long npages;
int ret;
int off;
int i;
DEFINE_DMA_ATTRS(attrs);
struct scatterlist *sg, *sg_list_start;
int need_release = 0;
if (dmasync)
dma_set_attr(DMA_ATTR_WRITE_BARRIER, &attrs);
@@ -97,7 +97,7 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
if (!can_do_mlock())
return ERR_PTR(-EPERM);
umem = kmalloc(sizeof *umem, GFP_KERNEL);
umem = kzalloc(sizeof *umem, GFP_KERNEL);
if (!umem)
return ERR_PTR(-ENOMEM);
@@ -117,8 +117,6 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
/* We assume the memory is from hugetlb until proved otherwise */
umem->hugetlb = 1;
INIT_LIST_HEAD(&umem->chunk_list);
page_list = (struct page **) __get_free_page(GFP_KERNEL);
if (!page_list) {
kfree(umem);
@@ -147,7 +145,18 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
cur_base = addr & PAGE_MASK;
ret = 0;
if (npages == 0) {
ret = -EINVAL;
goto out;
}
ret = sg_alloc_table(&umem->sg_head, npages, GFP_KERNEL);
if (ret)
goto out;
need_release = 1;
sg_list_start = umem->sg_head.sgl;
while (npages) {
ret = get_user_pages(current, current->mm, cur_base,
min_t(unsigned long, npages,
@@ -157,54 +166,38 @@ struct ib_umem *ib_umem_get(struct ib_ucontext *context, unsigned long addr,
if (ret < 0)
goto out;
umem->npages += ret;
cur_base += ret * PAGE_SIZE;
npages -= ret;
off = 0;
for_each_sg(sg_list_start, sg, ret, i) {
if (vma_list && !is_vm_hugetlb_page(vma_list[i]))
umem->hugetlb = 0;
while (ret) {
chunk = kmalloc(sizeof *chunk + sizeof (struct scatterlist) *
min_t(int, ret, IB_UMEM_MAX_PAGE_CHUNK),
GFP_KERNEL);
if (!chunk) {
ret = -ENOMEM;
goto out;
}
chunk->nents = min_t(int, ret, IB_UMEM_MAX_PAGE_CHUNK);
sg_init_table(chunk->page_list, chunk->nents);
for (i = 0; i < chunk->nents; ++i) {
if (vma_list &&
!is_vm_hugetlb_page(vma_list[i + off]))
umem->hugetlb = 0;
sg_set_page(&chunk->page_list[i], page_list[i + off], PAGE_SIZE, 0);
}
chunk->nmap = ib_dma_map_sg_attrs(context->device,
&chunk->page_list[0],
chunk->nents,
DMA_BIDIRECTIONAL,
&attrs);
if (chunk->nmap <= 0) {
for (i = 0; i < chunk->nents; ++i)
put_page(sg_page(&chunk->page_list[i]));
kfree(chunk);
ret = -ENOMEM;
goto out;
}
ret -= chunk->nents;
off += chunk->nents;
list_add_tail(&chunk->list, &umem->chunk_list);
sg_set_page(sg, page_list[i], PAGE_SIZE, 0);
}
ret = 0;
/* preparing for next loop */
sg_list_start = sg;
}
umem->nmap = ib_dma_map_sg_attrs(context->device,
umem->sg_head.sgl,
umem->npages,
DMA_BIDIRECTIONAL,
&attrs);
if (umem->nmap <= 0) {
ret = -ENOMEM;
goto out;
}
ret = 0;
out:
if (ret < 0) {
__ib_umem_release(context->device, umem, 0);
if (need_release)
__ib_umem_release(context->device, umem, 0);
kfree(umem);
} else
current->mm->pinned_vm = locked;
@@ -278,17 +271,16 @@ EXPORT_SYMBOL(ib_umem_release);
int ib_umem_page_count(struct ib_umem *umem)
{
struct ib_umem_chunk *chunk;
int shift;
int i;
int n;
struct scatterlist *sg;
shift = ilog2(umem->page_size);
n = 0;
list_for_each_entry(chunk, &umem->chunk_list, list)
for (i = 0; i < chunk->nmap; ++i)
n += sg_dma_len(&chunk->page_list[i]) >> shift;
for_each_sg(umem->sg_head.sgl, sg, umem->nmap, i)
n += sg_dma_len(sg) >> shift;
return n;
}