Warning - running *really* short on DMA buffers

[Mihai RUSU <dizzy@xxxxxxxxx>]

Hi

I am testing the 2.4.19-xfs_1.2pre3 kernel release. I compiled it from
sources with gcc 2.95.3. I am very glad to see much unlink() improvements
using version2 journal. Also external journal blasts away any internal
journal bench :) (external on another PCI bus, controller and disk).

I observed the message from the $subject in my kernel logs just right
after a mkfs.xfs line, here is what I used:
mkfs.xfs -f -l logdev=/dev/sda1,size=32768b,version=2 /dev/sdb1

/dev/sdb is a IBM ServeRAID
/dev/sda is a SCSI on-board controller/disk Adaptec 7899P (Intel SCB2
motherboard)

Looking in the sources I see this message printk() from
drivers/scsi/scsi_merge.c :

        /*
         * Now fill the scatter-gather table.
         */
        if (!sgpnt) {
                /*
                 * If we cannot allocate the scatter-gather table, then
                 * simply write the first buffer all by itself.
                 */
                printk("Warning - running *really* short on DMA buffers\n");
                this_count = SCpnt->request.current_nr_sectors;
                goto single_segment;
        }


What does that mean ? Where can I increase those "DMA buffers" ? What
could be the possible cause ? What does xfs.mkfs do that could trigger
that message ?

Thanks

----------------------------
Mihai RUSU