ide_register_hw(): buggy code

ide_register_hw(): buggy code

[Adrian Bunk]

The Coverity checker spotted the following bogus change to 
ide_register_hw() in commit 9e016a719209d95338e314b46c3012cc7feaaeec:

<--  snip  -->

...
+               hwif = ide_deprecated_find_port(hw->io_ports[IDE_DATA_OFFSET]);
+               index = hwif->index;
+               if (hwif)
+                       goto found;
                for (index = 0; index < MAX_HWIFS; index++)
...

<--  snip  -->

It's impossible to reach the for() loop without Oopsing before.

Can you either fix this for 2.6.25 or push your patch that removes 
ide_register_hw() for 2.6.25?

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed

Re: ide_register_hw(): buggy code

[Peter Teoh]

On Sun, Mar 2, 2008 at 11:19 PM, Adrian Bunk <bunk@kernel.org> wrote:
> The Coverity checker spotted the following bogus change to
>  ide_register_hw() in commit 9e016a719209d95338e314b46c3012cc7feaaeec:
>
>  <--  snip  -->
>
>  ...
>  +               hwif = ide_deprecated_find_port(hw->io_ports[IDE_DATA_OFFSET]);
>  +               index = hwif->index;
>  +               if (hwif)
>  +                       goto found;
>                 for (index = 0; index < MAX_HWIFS; index++)
>  ...
>
>  <--  snip  -->
>
>  It's impossible to reach the for() loop without Oopsing before.
>
>  Can you either fix this for 2.6.25 or push your patch that removes
>  ide_register_hw() for 2.6.25?
>

My question is:

a.   why is "retry=1", and then the do while loop always end up the
loop being one round executed only?   Why not just remove the while
loop entirely?

b.   not sure if your statement above implied this, but checking for
hwif!=0 should be before index.  ???

c.   "index = hwif->index;" should not be there, but after "found".
Is that correct?

int ide_register_hw(hw_regs_t *hw, void (*quirkproc)(ide_drive_t *),
                    ide_hwif_t **hwifp)
{
        int index, retry = 1;
        ide_hwif_t *hwif;
        u8 idx[4] = { 0xff, 0xff, 0xff, 0xff };

        do {
                hwif = ide_deprecated_find_port(hw->io_ports[IDE_DATA_OFFSET]);
                index = hwif->index;
                if (hwif)
                        goto found;
                for (index = 0; index < MAX_HWIFS; index++)
                        ide_unregister(index, 1, 1);
        } while (retry--);
        return -1;
found:
        if (hwif->present)


The patch I am thinking goes something like this:

--- ide/ide.c   2008-03-03 20:10:28.000000000 +0800
+++ ide/ide_new.c       2008-03-04 00:09:46.000000000 +0800
@@ -661,20 +661,18 @@ EXPORT_SYMBOL_GPL(ide_deprecated_find_po
 int ide_register_hw(hw_regs_t *hw, void (*quirkproc)(ide_drive_t *),
                    ide_hwif_t **hwifp)
 {
-       int index, retry = 1;
+       int index;
        ide_hwif_t *hwif;
        u8 idx[4] = { 0xff, 0xff, 0xff, 0xff };

-       do {
-               hwif = ide_deprecated_find_port(hw->io_ports[IDE_DATA_OFFSET]);
-               index = hwif->index;
-               if (hwif)
-                       goto found;
-               for (index = 0; index < MAX_HWIFS; index++)
-                       ide_unregister(index, 1, 1);
-       } while (retry--);
+       hwif = ide_deprecated_find_port(hw->io_ports[IDE_DATA_OFFSET]);
+       if (hwif)
+               goto found;
+       for (index = 0; index < MAX_HWIFS; index++)
+               ide_unregister(index, 1, 1);
        return -1;
 found:
+       index = hwif->index;
        if (hwif->present)
                ide_unregister(index, 0, 1);
        else if (!hwif->hold)

Please comment.

-- 
Regards,
Peter Teoh

Re: ide_register_hw(): buggy code

[Bartlomiej Zolnierkiewicz]

Hi,

On Monday 03 March 2008, Peter Teoh wrote:
> On Sun, Mar 2, 2008 at 11:19 PM, Adrian Bunk <bunk@kernel.org> wrote:
> > The Coverity checker spotted the following bogus change to
> >  ide_register_hw() in commit 9e016a719209d95338e314b46c3012cc7feaaeec:
> >
> >  <--  snip  -->
> >
> >  ...
> >  +               hwif = ide_deprecated_find_port(hw->io_ports[IDE_DATA_OFFSET]);
> >  +               index = hwif->index;
> >  +               if (hwif)
> >  +                       goto found;
> >                 for (index = 0; index < MAX_HWIFS; index++)
> >  ...
> >
> >  <--  snip  -->
> >
> >  It's impossible to reach the for() loop without Oopsing before.

[ iff free hwif is not found (unlikely case) ]

> >  Can you either fix this for 2.6.25 or push your patch that removes
> >  ide_register_hw() for 2.6.25?
> >
> 
> My question is:
> 
> a.   why is "retry=1", and then the do while loop always end up the
> loop being one round executed only?   Why not just remove the while
> loop entirely?

the whole ide_register_hw() is already gone in IDE tree
(these patches are scheduled for 2.6.26)

> b.   not sure if your statement above implied this, but checking for
> hwif!=0 should be before index.  ???
> 
> c.   "index = hwif->index;" should not be there, but after "found".
> Is that correct?

Yes, could you please re-do your patch to contain:

- only 'hwif->index' change
- proper patch description
- Signed-off-by: line

so I could merge it?

Thanks,
Bart

Re: ide_register_hw(): buggy code

[Peter Teoh]

[-- Attachment #1: Type: text/plain, Size: 1791 bytes --]

On Tue, Mar 4, 2008 at 6:29 AM, Bartlomiej Zolnierkiewicz
<bzolnier@gmail.com> wrote:
>
>  Hi,
>
>
>  On Monday 03 March 2008, Peter Teoh wrote:
>  > On Sun, Mar 2, 2008 at 11:19 PM, Adrian Bunk <bunk@kernel.org> wrote:
>  > > The Coverity checker spotted the following bogus change to
>  > >  ide_register_hw() in commit 9e016a719209d95338e314b46c3012cc7feaaeec:
>  > >
>  > >  <--  snip  -->
>  > >
>  > >  ...
>  > >  +               hwif = ide_deprecated_find_port(hw->io_ports[IDE_DATA_OFFSET]);
>  > >  +               index = hwif->index;
>  > >  +               if (hwif)
>  > >  +                       goto found;
>  > >                 for (index = 0; index < MAX_HWIFS; index++)
>  > >  ...
>  > >
>  > >  <--  snip  -->
>  > >
>  > >  It's impossible to reach the for() loop without Oopsing before.
>
>  [ iff free hwif is not found (unlikely case) ]
>
>
>  > >  Can you either fix this for 2.6.25 or push your patch that removes
>  > >  ide_register_hw() for 2.6.25?
>  > >
>  >
>  > My question is:
>  >
>  > a.   why is "retry=1", and then the do while loop always end up the
>  > loop being one round executed only?   Why not just remove the while
>  > loop entirely?
>
>  the whole ide_register_hw() is already gone in IDE tree
>  (these patches are scheduled for 2.6.26)
>
>
>  > b.   not sure if your statement above implied this, but checking for
>  > hwif!=0 should be before index.  ???
>  >
>  > c.   "index = hwif->index;" should not be there, but after "found".
>  > Is that correct?
>
>  Yes, could you please re-do your patch to contain:
>
>  - only 'hwif->index' change
>  - proper patch description
>  - Signed-off-by: line
>
>  so I could merge it?


Description:

Relocating the index to come after finding the hwif pointer.

Thanks.

-- 
Regards,
Peter Teoh

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: relocating_deriving_index.patch --]
[-- Type: text/x-patch; name=relocating_deriving_index.patch, Size: 614 bytes --]

Signed-off-by: Peter Teoh <htmldeveloper@gmail.com>

--- drivers/ide/ide.c.orig	2008-03-04 08:26:11.000000000 +0800
+++ drivers/ide/ide.c	2008-03-04 09:07:44.000000000 +0800
@@ -667,7 +667,6 @@ int ide_register_hw(hw_regs_t *hw, void 
 
 	do {
 		hwif = ide_deprecated_find_port(hw->io_ports[IDE_DATA_OFFSET]);
-		index = hwif->index;
 		if (hwif)
 			goto found;
 		for (index = 0; index < MAX_HWIFS; index++)
@@ -675,6 +674,7 @@ int ide_register_hw(hw_regs_t *hw, void 
 	} while (retry--);
 	return -1;
 found:
+	index = hwif->index;
 	if (hwif->present)
 		ide_unregister(index, 0, 1);
 	else if (!hwif->hold)

Re: ide_register_hw(): buggy code

[Bartlomiej Zolnierkiewicz]

On Tuesday 04 March 2008, Peter Teoh wrote:
> On Tue, Mar 4, 2008 at 6:29 AM, Bartlomiej Zolnierkiewicz
> <bzolnier@gmail.com> wrote:
> >
> >  Hi,
> >
> >
> >  On Monday 03 March 2008, Peter Teoh wrote:
> >  > On Sun, Mar 2, 2008 at 11:19 PM, Adrian Bunk <bunk@kernel.org> wrote:
> >  > > The Coverity checker spotted the following bogus change to
> >  > >  ide_register_hw() in commit 9e016a719209d95338e314b46c3012cc7feaaeec:
> >  > >
> >  > >  <--  snip  -->
> >  > >
> >  > >  ...
> >  > >  +               hwif = ide_deprecated_find_port(hw->io_ports[IDE_DATA_OFFSET]);
> >  > >  +               index = hwif->index;
> >  > >  +               if (hwif)
> >  > >  +                       goto found;
> >  > >                 for (index = 0; index < MAX_HWIFS; index++)
> >  > >  ...
> >  > >
> >  > >  <--  snip  -->
> >  > >
> >  > >  It's impossible to reach the for() loop without Oopsing before.
> >
> >  [ iff free hwif is not found (unlikely case) ]
> >
> >
> >  > >  Can you either fix this for 2.6.25 or push your patch that removes
> >  > >  ide_register_hw() for 2.6.25?
> >  > >
> >  >
> >  > My question is:
> >  >
> >  > a.   why is "retry=1", and then the do while loop always end up the
> >  > loop being one round executed only?   Why not just remove the while
> >  > loop entirely?
> >
> >  the whole ide_register_hw() is already gone in IDE tree
> >  (these patches are scheduled for 2.6.26)
> >
> >
> >  > b.   not sure if your statement above implied this, but checking for
> >  > hwif!=0 should be before index.  ???
> >  >
> >  > c.   "index = hwif->index;" should not be there, but after "found".
> >  > Is that correct?
> >
> >  Yes, could you please re-do your patch to contain:
> >
> >  - only 'hwif->index' change
> >  - proper patch description
> >  - Signed-off-by: line
> >
> >  so I could merge it?
> 
> 
> Description:
> 
> Relocating the index to come after finding the hwif pointer.

applied, thanks