[PATCH] Fix get_graphics_device() - returns error if graphics tag is missing
# HG changeset patch # User Kaitlin Rupert <karupert@us.ibm.com> # Date 1199734960 28800 # Node ID abcb596636c9530009756b1aa1e8900a2f01d10e # Parent ea65740aa1ffe73b47a4a14c8a853dbb3f040016 Fix get_graphics_device() - returns error if graphics tag is missing. The graphics tag isn't mandatory, so we shouldn't return an error here. The same problem exists with get_emu_device(). Signed-off-by: Kaitlin Rupert <karupert@us.ibm.com> diff -r ea65740aa1ff -r abcb596636c9 libxkutil/device_parsing.c --- a/libxkutil/device_parsing.c Fri Jan 04 12:53:44 2008 -0800 +++ b/libxkutil/device_parsing.c Mon Jan 07 11:42:40 2008 -0800 @@ -395,7 +395,7 @@ static int get_emu_device(virDomainPtr d free(xml); - return ret; + return 1; } static int get_graphics_device(virDomainPtr dom, struct virt_device **dev) @@ -416,7 +416,7 @@ static int get_graphics_device(virDomain free(xml); - return ret; + return 1; } int get_disk_devices(virDomainPtr dom, struct virt_device **list)
Kaitlin Rupert wrote:
# HG changeset patch # User Kaitlin Rupert <karupert@us.ibm.com> # Date 1199734960 28800 # Node ID abcb596636c9530009756b1aa1e8900a2f01d10e # Parent ea65740aa1ffe73b47a4a14c8a853dbb3f040016 Fix get_graphics_device() - returns error if graphics tag is missing.
The graphics tag isn't mandatory, so we shouldn't return an error here. The same problem exists with get_emu_device().
Signed-off-by: Kaitlin Rupert <karupert@us.ibm.com>
diff -r ea65740aa1ff -r abcb596636c9 libxkutil/device_parsing.c --- a/libxkutil/device_parsing.c Fri Jan 04 12:53:44 2008 -0800 +++ b/libxkutil/device_parsing.c Mon Jan 07 11:42:40 2008 -0800 @@ -395,7 +395,7 @@ static int get_emu_device(virDomainPtr d
free(xml);
- return ret; + return 1; }
static int get_graphics_device(virDomainPtr dom, struct virt_device **dev) @@ -416,7 +416,7 @@ static int get_graphics_device(virDomain
free(xml);
- return ret; + return 1; }
int get_disk_devices(virDomainPtr dom, struct virt_device **list)
If the return code of these two functions does not matter, couldn't we also make a void function out of them ? This could avoid the impression that the result of these functions matters for the return code of the calling function, like in get_dominfo(). -- Regards Heidi Eckhart Software Engineer IBM Linux Technology Center - Open Hypervisor
KR> - return ret; KR> + return 1; This will cause the do_parse() function to keep an empty (and uninitialized) device struct on the list, because of this:
if (do_real_parse(dev_nodes[devidx], &list[lstidx])) lstidx++;
Which is bad. I think we need to keep the conditional return value From those individual functions. I think the problem lies just with this bit in get_dominfo():
ret = get_emu_device(dom, &(*dominfo)->dev_emu); ret = get_graphics_device(dom, &(*dominfo)->dev_graphics);
Which was me thinking that I'd capture the return value and maybe check it later, but ignore for now. The problem is that we return ret a few lines down, which makes the function fail if one of the two above calls fail. So, I think the solution here for now is to either do something meaningful with ret (not sure what that would be), or just remove 'ret =' from each call and truly ignore the status of those. Also, KR> - return ret; KR> + return 1; These functions now return a bool instead of an int. Maybe you need to update your tree? Thanks! -- Dan Smith IBM Linux Technology Center Open Hypervisor Team email: danms@us.ibm.com
Dan Smith wrote:
KR> - return ret; KR> + return 1;
This will cause the do_parse() function to keep an empty (and uninitialized) device struct on the list, because of this:
if (do_real_parse(dev_nodes[devidx], &list[lstidx])) lstidx++;
Ah good point. Yep, I didn't notice this.
Which is bad. I think we need to keep the conditional return value From those individual functions. I think the problem lies just with this bit in get_dominfo():
ret = get_emu_device(dom, &(*dominfo)->dev_emu); ret = get_graphics_device(dom, &(*dominfo)->dev_graphics);
Which was me thinking that I'd capture the return value and maybe check it later, but ignore for now. The problem is that we return ret a few lines down, which makes the function fail if one of the two above calls fail.
So, I think the solution here for now is to either do something meaningful with ret (not sure what that would be), or just remove 'ret =' from each call and truly ignore the status of those.
I'd vote for not capturing the return from the get_emu_device() and get_graphics_device() in get_dominfo(). Trying to handle the return in a meaningful way would probably not be obvious and would need a comment.. -- Kaitlin Rupert IBM Linux Technology Center karupert@us.ibm.com
participants (3)
-
Dan Smith -
Heidi Eckhart -
Kaitlin Rupert