On Wed, Sep 4, 2024 at 3:26 AM Jeff Layton <jlayton@xxxxxxxxxx> wrote: > > On Tue, 2024-09-03 at 13:32 -0700, Rick Macklem wrote: > > Hi, > > > > Subject line says it all. As a newbie to the Linux > > kernel, I am wondering if an array of 70 pointers > > is too big for the kernel stack? > > (I assumed it is and kmalloc_array()d it, but thought > > I'd check.) > > > > Thanks for any comments, rick > > > > In the old days, we had 4k stacks and it was quite easy to blow it out > putting big stuff there. These days, stacks are 16k (on most arches) so > you're probably fine with a 560 byte array there in most cases. > > Still, with something that large, I usually do a kmalloc, just to be > sure. Yea, I was thinking the same thing. And then I came across this comment in the NFSv3 acl code... /* Avoid the use of posix_acl_alloc(). nfsacl_encode() is * invoked in contexts where a memory allocation failure is * fatal. Fortunately this fake ACL is small enough to * construct on the stack. */ So, now I have the array on the stack and it seems to work. (Mind you, I am testing on 32bit, so I'm not sure what will happen for others. Yea, believe it or not, my only Linux box is i686.;-( Thanks for the help, rick > -- > Jeff Layton <jlayton@xxxxxxxxxx>
