Re: [Patch 3/4] Memory-mapped writing and reading of a graded file spanning over two media types

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 04/06/2018 04:41 AM, Sayan Ghosh wrote:
> The patch is on top of Linux Kernel 4.7.2.
> 
> Signed-off-by: Sayan Ghosh <sgdgp.2014@xxxxxxxxx>
> ---
>  fs/ext4/file.c | 178 +++++++++++++++++++++++++++++++++++++++++++++++++++++++--
>  1 file changed, 174 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/ext4/file.c b/fs/ext4/file.c
> index df44c87..368cf53 100755
> --- a/fs/ext4/file.c
> +++ b/fs/ext4/file.c
> @@ -298,6 +298,164 @@ static const struct vm_operations_struct
> ext4_file_vm_ops = {
>      .page_mkwrite   = ext4_page_mkwrite,
>  };
> 
> +/*
> + * This function is the fault function for our case to
> + * redirect the high grade blocks through DAX path (since the
> + * higher tier we chose is Persistent Memory) and the lower
> + * grade blocks via the normal ext4 fault.
> + * Some parts of the code are copied from ext4_dax_fault.
> + * The parts where the redirection is done is added by additional
> + * comments.
> + */
> +static int graded_ext4_fault(struct vm_area_struct *vma, struct vm_fault *vmf){

Put '{' on separate line.

> +    int result;
> +    sector_t block;
> +    handle_t *handle = NULL;
> +    struct file *file = vma->vm_file;
> +    struct address_space *mapping = file->f_mapping;
> +    struct inode *inode = file_inode(vma->vm_file);
> +    struct super_block *sb = inode->i_sb;
> +    bool write = vmf->flags & FAULT_FLAG_WRITE;

Indentation. (many)

> +
> +    struct grade_struct *grade_array = NULL;
> +    unsigned long long total;
> +    if (is_file_graded(inode)){
> +        total = read_count_xattr(inode);
> +        grade_array = (struct grade_struct
> *)kmalloc(total*sizeof(struct grade_struct), GFP_USER);
> +        read_grade_xattr(inode,grade_array);
> +    }
> +
> +    block = (sector_t)vmf->pgoff << (PAGE_SHIFT - mapping->host->i_blkbits);
> +    if (write) {
> +        sb_start_pagefault(sb);
> +        file_update_time(vma->vm_file);
> +        down_read(&EXT4_I(inode)->i_mmap_sem);
> +        handle = ext4_journal_start_sb(sb, EXT4_HT_WRITE_PAGE,
> +                        EXT4_DATA_TRANS_BLOCKS(sb));
> +    } else
> +        down_read(&EXT4_I(inode)->i_mmap_sem);
> +
> +    if (IS_ERR(handle))
> +        result = VM_FAULT_SIGBUS;
> +    else
> +    {
> +        if(write){

		if (write) {

> +            unsigned long long temp;
> +            if(find_grade(grade_array,total,block,&temp) == 1){

		if (                                           1) {

> +                result = __dax_fault(vma, vmf, ext4_dax_get_block);
> +            }
> +            else if(find_grade(grade_array,total,block,&temp) == 0){

same.

> +                result = ext4_filemap_fault(vma,vmf);
> +            }
> +        }
> +        else{

	else {

> +            /*
> +             * Here the higher graded blocks are redirected via DAX path
> +             * since we consider Persistent Memory as higher tier.
> +             *
> +             * ** TODO **
> +             * To take care of the case when the higher tier is not
> +             * persistent memory (can be HDD-SSD combination), a check
> +             * of the same needs to be provided before re-direction.
> +             */
> +            unsigned long long temp;
> +            if(find_grade(grade_array,total,block,&temp) == 1){
> +                result = __dax_fault(vma, vmf, ext4_dax_get_block);
> +            }
> +            else if(find_grade(grade_array,total,block,&temp) == 0){
> +                result = ext4_filemap_fault(vma,vmf);
> +            }
> +        }
> +    }
> + out:
> +    if (write) {
> +        if (!IS_ERR(handle))
> +            ext4_journal_stop(handle);
> +        up_read(&EXT4_I(inode)->i_mmap_sem);
> +        sb_end_pagefault(sb);
> +    } else
> +        up_read(&EXT4_I(inode)->i_mmap_sem);
> +
> +    return result;
> +}
> +
> +/*
> + * This is the new page write function for our scenario.
> + * This also takes care of the grade and redirects
> + * through the correct path, DAX for higer tier
> + * (Persistent Memory) and ext4 path for lower tier.
> + * To take care of the cases when the higher tier
> + * is not Persistent Memory a TODO has been added
> + */
> +static int graded_ext4_mkwrite(struct vm_area_struct *vma, struct
> vm_fault *vmf){

{ on separate line.

> +    int result;
> +    sector_t block;
> +    handle_t *handle = NULL;
> +    struct file *file = vma->vm_file;
> +    struct address_space *mapping = file->f_mapping;
> +    struct inode *inode = file_inode(vma->vm_file);
> +    struct super_block *sb = inode->i_sb;
> +    bool write = vmf->flags & FAULT_FLAG_WRITE;
> +    block = (sector_t)vmf->pgoff << (PAGE_SHIFT - mapping->host->i_blkbits);
> +
> +    struct grade_struct *grade_array = NULL;
> +    unsigned long long total;
> +    if (is_file_graded(inode)){
> +        total = read_count_xattr(inode);
> +        grade_array = (struct grade_struct
> *)kmalloc(total*sizeof(struct grade_struct), GFP_USER);
> +        read_grade_xattr(inode,grade_array);
> +    }
> +
> +    if (write) {
> +        sb_start_pagefault(sb);
> +        file_update_time(vma->vm_file);
> +        down_read(&EXT4_I(inode)->i_mmap_sem);
> +        handle = ext4_journal_start_sb(sb, EXT4_HT_WRITE_PAGE,
> +                        EXT4_DATA_TRANS_BLOCKS(sb));
> +    } else
> +        down_read(&EXT4_I(inode)->i_mmap_sem);
> +
> +    if (IS_ERR(handle))
> +        result = VM_FAULT_SIGBUS;
> +    else{

	else {

> +          /*
> +         * Here the higher graded blocks are redirected via DAX path
> +         * since we consider Persistent Memory as higher tier.
> +         *
> +         * ** TODO **
> +         * To take care of the case when the higher tier is not
> +         * persistent memory (can be HDD-SSD combination), a check
> +         * of the same needs to be provided before re-direction.
> +         */
> +          unsigned long long temp;
> +        if(find_grade(grade_array,total,block,&temp)==1){
> +            result = __dax_fault(vma, vmf, ext4_dax_get_block);
> +        }
> +        else if(find_grade(grade_array,total,block,&temp)==0){
> +            filemap_map_pages(vma,vmf);
> +            result = ext4_page_mkwrite(vma,vmf);
> +        }
> +    }
> +    if (write) {
> +        if (!IS_ERR(handle))
> +            ext4_journal_stop(handle);
> +        up_read(&EXT4_I(inode)->i_mmap_sem);
> +        sb_end_pagefault(sb);
> +    } else
> +        up_read(&EXT4_I(inode)->i_mmap_sem);
> +
> +    return result;
> +}
> +
> +/*
> + * New function pointers for page fault handling and page writes.
> + */
> +static const struct vm_operations_struct graded_ext4_vm_ops = {
> +    .fault = graded_ext4_fault,
> +    .page_mkwrite = graded_ext4_mkwrite,
> +
> +};
> +
>  static int ext4_file_mmap(struct file *file, struct vm_area_struct *vma)
>  {
>      struct inode *inode = file->f_mapping->host;
> @@ -310,11 +468,23 @@ static int ext4_file_mmap(struct file *file,
> struct vm_area_struct *vma)
>              return -ENOKEY;
>      }
>      file_accessed(file);
> -    if (IS_DAX(file_inode(file))) {
> -        vma->vm_ops = &ext4_dax_vm_ops;
> +
> +    /*
> +     * For graded file new function pointers for
> +     * fault and page write are assigned.
> +     */
> +    if(is_file_graded(file_inode(file))){

	if (                            ))) {

> +        vma->vm_ops = &graded_ext4_vm_ops;
>          vma->vm_flags |= VM_MIXEDMAP | VM_HUGEPAGE;
> -    } else {
> -        vma->vm_ops = &ext4_file_vm_ops;
> +    }
> +    else{

	else {

> +        if (IS_DAX(file_inode(file))) {
> +            vma->vm_ops = &ext4_dax_vm_ops;
> +            vma->vm_flags |= VM_MIXEDMAP | VM_HUGEPAGE;
> +        }
> +        else {
> +            vma->vm_ops = &ext4_file_vm_ops;
> +        }
>      }
>      return 0;
>  }
> ‌
> 


-- 
~Randy



[Index of Archives]     [Linux Ext4 Filesystem]     [Union Filesystem]     [Filesystem Testing]     [Ceph Users]     [Ecryptfs]     [AutoFS]     [Kernel Newbies]     [Share Photos]     [Security]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux Cachefs]     [Reiser Filesystem]     [Linux RAID]     [Samba]     [Device Mapper]     [CEPH Development]

  Powered by Linux