How awful a practice is it to use symlinks to kernel
modules when a matching libafs is not found for the
*exact* kernel?
That is, for instance:
libafs-2.6.18-128.1.6.el5.mp.ko -> libafs-2.6.18-92.el5.mp.ko
libafs-2.6.18-164.11.1.el5.mp.ko -> libafs-2.6.18-92.el5.mp.ko
libafs-2.6.18-92.el5.mp.ko
It's not without risk. You're probably ok (and assuming the symbol
versioning works correctly you shouldn't lose if the module does load)
but we don't want to rely on it.
On Tue, Jan 26, 2010 at 12:08 PM, Jeff Blaine wrote:
> How awful a practice is it to use symlinks to kernel
> modules when a m
On 26 Jan 2010, at 17:11, Derrick Brashear wrote:
> It's not without risk. You're probably ok (and assuming the symbol
> versioning works correctly you shouldn't lose if the module does load)
> but we don't want to rely on it.
RedHat already do this with weak-updates. In theory they guarantee th
Simon Wilkinson writes:
> On 26 Jan 2010, at 17:11, Derrick Brashear wrote:
>
>> It's not without risk. You're probably ok (and assuming the symbol
>> versioning works correctly you shouldn't lose if the module does load)
>> but we don't want to rely on it.
>
> RedHat already do this with weak-up
