Errata for Linux Device Drivers

root# rmmod hello.o now reads: root# rmmod hello

in the last line of the paragraph change "the module defines the execution" to "the iBCS2 module defines the execution"

and line 2: "functions, thus why" changed to "functions. That's why"

The code used to read: return filp->f_op->open(inode, filp); It now reads: return filp->f_op->open(inode, filp); /* dispatch to specific open */

count = quantum - q_pos; /* read only up to /* to count = quantum - q_pos; /* read only up to */

count = quantum -q_pos; /* write only up to /* to count = quantum -q_pos; /* write only up to */

Where it says: "issue the command cat /proc/kmesg" it should say: "issue the command cat /proc/kmsg"

"Another advantage of oops is ..." with Like the original ksyms, the program is able to decode oops messages generated on Intel platforms only. Porting to other platforms shouldn't be difficult. ("ksysm" is in italics.)

"(the next queue in the list)" to "(the next item in the list)"

"since the computer was turned on" to "since the operating system was booted"

Mr Rubini introduced the "Digital Resistor" to us. If resistors could store data!? Please replace "1KB" resistor with {LaTeX} "1k/ohmega resistor"

"the self object" to "the this object" ("this" is in constant-width font)

"Going to Sleep Without Races", the code had a subtle bug. The correct implementation of the trick looks like the following: add_wait_queue(&short_queue, &wait); do { current->state = TASK_INTERRUPTIBLE; schedule(); } while ((short_head == short_tail) && !(current->signal & ~current->blocked) ); remove_wait_queue(&short_queue, &wait); if (current->signal & ~current->blocked) /* a signal arrived */ return -ERESTARTSYS; This is different from the version shown in the printed text as signal decoding must be performed *after* calling remove_wait_queue().

("export.ver" is italics)

"end_request(0)" (in constant-width font)

following: int (*swapout)(struct vm_area_struct *vma, unsigned long offset, pte_t *page_table); This method is called when a page is selected for swap-out. The [cw]offset[/] argument is the file offset: [cw]virt_address - vma->vm_start + vma->vm_offset[/]. A return value of 0 signals success; any other value signals an error. In case of error, the process owning the page is sent a [cw]SIGBUS[/]. In version 1.2, the function returned [cw]void[/] because it was never expected to fail. The method usually writes useful information to [cw]*page_table[/], so it can be retrieved at swap-in. Such information can be, for example, the offset in the swap device. pte_t (*swapin)(struct vm_area_struct *, unsigned long offset, unsigned long entry); This method is used to retrieve a page from swap space. The [cw]offset[/] argument is relative to the area (as above for <F>swapout</>), while [cw]entry[/] is the current [cw]pte[/] for the page -- if <F>swapout</> saved some information in the entry, that information can now be used to retrieve the page. ("[cw]" and "[/]" indicate that text that should appear in constant width. "<F>" and "</>" indicate that text that should appear in italic.)

"the X server transfers a lot of data to and from video memory." to "the X server transfers a lot of data from video memory."

The last sentence on the page now reads: "If you use kmalloc you must specify GFP_DMA priority, ...."

another way to allocate" to the end of the section. In their place, added the following paragraph: My own implementation of allocating DMA buffers is available as an allocator.c module (with an accompanying header). You can find a version in the sample files within src/misc-modules; and the latest version is always available from my own ftp site, ftp://ftp.systemy.it/pub/develop. You could also look for the bigphysarea kernel patch, which is meant to accomplish the same goal as my allocator. ("allocator.c," "src/misc-modules," "ftp://ftp.systemy.it/pub/ develop" and "big physarea" are all in italics.)

look better after we removed three paragraphs from the preceding section.) Added the following paragraph and code sample after the paragraphh ending, "...when address information is received from hardware connected to the bus." If you look at code that relies on the allocator engine described earlier, you'll find a sample use of these functions. The code relies also on vremap because of the context: ("allocator" and "vremap" are in italics.) /* the allocator returns a physical address */ dev->dmabuffer = allocator_allocate_dma(kilobytes, GFP_KERNEL); /* map it in the virtual address space */ dev->dmavbuffer = vremap(dev-&gt;dmabuffer, kilobytes*1024); /* ... */ /* pass the address to the device */ writel(virt_to_bus(dev->dmavbuffer, dev->registers + DEV_DMA_ADDRESS);

"memcpy_tofs" was changed to: "copy_to_user".

The text in the middle of the page used to read: "memcpy_tofs" It now reads: "memcpy_fromfs"

if (_IOC_DIR(cmd) & _IOC_READ) { if (!access_ok(VERIFY_WRITE, (void *)arg, size)) retturn -EFAULT; "retturn" should be "return"

Interface", the ioctl() implementation shown included a mistake: "retval" is not initialized. A correct implementation is: int another_ioctl(struct inode *inode, struct file *filp, unsigned int cmd, unsigned long arg); { int retval = -EINVAL, size = _IOC_SIZE(cmd); if (_IOC_DIR(cmd) & _IOC_READ) { if (!access_ok(VERIFY_WRITE, (void *)arg, size)) retturn -EFAULT; } else if (_IOC_DIR(cmd) & _IOC_WRITE) { if (!access_ok(VERIFY_READ, (void *)arg, size)) return -EFAULT switch(cmd) { case NEW_GETVALUE: retval = __put_user(another_value,arg); break; case NEW_SETVALUE: retval = __get_user(another_value,arg); break; } return retval; }

"16-byte little-endian" to "16-bit little-endian"

"The second pointer of the table" to "The second pointer of each pair in the table"