Re: [PATCH v13 39/45] sg: add mmap_sz tracking

[Douglas Gilbert <dgilbert@interlog.com>]

On 2021-01-14 2:18 a.m., Hannes Reinecke wrote:
> On 1/13/21 11:45 PM, Douglas Gilbert wrote:
>> Track mmap_sz from prior mmap(2) call, per sg file descriptor. Also
>> reset this value whenever munmap(2) is called. Fail SG_FLAG_MMAP_IO
>> uses if mmap(2) hasn't been called or the memory associated with it
>> is not large enough for the current request.
>>
>> Signed-off-by: Douglas Gilbert <dgilbert@interlog.com>
>> ---
>>   drivers/scsi/sg.c | 11 +++++++++--
>>   1 file changed, 9 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/scsi/sg.c b/drivers/scsi/sg.c
>> index c5533bbaf0d5..d0f25f8f572b 100644
>> --- a/drivers/scsi/sg.c
>> +++ b/drivers/scsi/sg.c
>> @@ -232,6 +232,7 @@ struct sg_fd {        /* holds the state of a file 
>> descriptor */
>>       atomic_t waiting;    /* number of requests awaiting receive */
>>       atomic_t req_cnt;    /* number of requests */
>>       int sgat_elem_sz;    /* initialized to scatter_elem_sz */
>> +    int mmap_sz;        /* byte size of previous mmap() call */
>>       unsigned long ffd_bm[1];    /* see SG_FFD_* defines above */
>>       pid_t tid;        /* thread id when opened */
>>       u8 next_cmd_len;    /* 0: automatic, >0: use on next write() */
>> @@ -725,10 +726,14 @@ sg_write(struct file *filp, const char __user *p, size_t 
>> count, loff_t *ppos)
>>   static inline int
>>   sg_chk_mmap(struct sg_fd *sfp, int rq_flags, int len)
>>   {
>> +    if (!test_bit(SG_FFD_MMAP_CALLED, sfp->ffd_bm))
>> +        return -EBADFD;
>>       if (atomic_read(&sfp->submitted) > 0)
>>           return -EBUSY;  /* already active requests on fd */
>>       if (len > sfp->rsv_srp->sgat_h.buflen)
>>           return -ENOMEM; /* MMAP_IO size must fit in reserve */
>> +    if (unlikely(len > sfp->mmap_sz))
>> +        return -ENOMEM; /* MMAP_IO size can't exceed mmap() size */
>>       if (rq_flags & SG_FLAG_DIRECT_IO)
>>           return -EINVAL; /* not both MMAP_IO and DIRECT_IO */
>>       return 0;
>> @@ -2262,6 +2267,8 @@ sg_vma_close(struct vm_area_struct *vma)
>>           pr_warn("%s: sfp null\n", __func__);
>>           return;
>>       }
>> +    sfp->mmap_sz = 0;
>> +    clear_bit(SG_FFD_MMAP_CALLED, sfp->ffd_bm);
>>       kref_put(&sfp->f_ref, sg_remove_sfp); /* get in: sg_vma_open() */
>>   }
>> @@ -2383,6 +2390,7 @@ sg_mmap(struct file *filp, struct vm_area_struct *vma)
>>               goto fini;
>>           }
>>       }
>> +    sfp->mmap_sz = req_sz;
>>       vma->vm_flags |= VM_IO | VM_DONTEXPAND | VM_DONTDUMP;
>>       vma->vm_private_data = sfp;
>>       vma->vm_ops = &sg_mmap_vm_ops;
>> @@ -4048,8 +4056,7 @@ sg_proc_debug_fd(struct sg_fd *fp, char *obp, int len, 
>> unsigned long idx)
>>                  (int)test_bit(SG_FFD_FORCE_PACKID, fp->ffd_bm),
>>                  (int)test_bit(SG_FFD_KEEP_ORPHAN, fp->ffd_bm),
>>                  fp->ffd_bm[0]);
>> -    n += scnprintf(obp + n, len - n, "   mmap_called=%d\n",
>> -               test_bit(SG_FFD_MMAP_CALLED, fp->ffd_bm));
>> +    n += scnprintf(obp + n, len - n, "   mmap_sz=%d\n", fp->mmap_sz);
>>       n += scnprintf(obp + n, len - n,
>>                  "   submitted=%d waiting=%d   open thr_id=%d\n",
>>                  atomic_read(&fp->submitted),
>>
> What is the point of having the SG_FFD_MMAP_CALLED bit now?
> Doesn't the new 'mmap_sz' value carry the same information?

Touché! Will fix.

Doug Gilbert