On 29.02.24 17:31, Jan Beulich wrote:
On 29.02.2024 17:29, Jürgen Groß wrote:On 29.02.24 16:46, Jan Beulich wrote:On 12.12.2023 10:47, Juergen Gross wrote:Allow 16 bits per cpu number, which is the limit imposed by spinlock_tickets_t.This will allow up to 65535 cpus, while increasing only the size of recursive spinlocks in debug builds from 8 to 12 bytes.I think we want to be more conservative here, for the case of there being bugs: The CPU holding a lock may wrongly try to acquire it a 2nd time. That's the 65536th ticket then, wrapping the value.Is this really a problem? There will be no other cpu left seeing the lock as "free" in this case, as all others will be waiting for the head to reach their private tail value.But isn't said CPU then going to make progress, rather than indefinitely spinning on the lock?
No, I don't think so. The limit isn't 65535 because of the ticket mechanism, but because of the rspin mechanism, where we need a "no cpu is owning the lock" value. Without the recursive locks the limit would be 65536 (or 4096 today).
Therefore my suggestion would be to only (mention) go(ing) up to 32k.Signed-off-by: Juergen Gross <jgr...@suse.com> --- xen/common/spinlock.c | 1 + xen/include/xen/spinlock.h | 18 +++++++++--------- 2 files changed, 10 insertions(+), 9 deletions(-)Shouldn't this also bump the upper bound of the NR_CPUS range then in xen/arch/Kconfig?Fine with me, I can add another patch to the series doing that.Why not do it right here? The upper bound there is like it is only because of the restriction that's lifted here.
I'd prefer splitting the two instances, but if you prefer it to be in a single patch, so be it. Juergen
OpenPGP_0xB0DE9DD628BF132F.asc
Description: OpenPGP public key
OpenPGP_signature.asc
Description: OpenPGP digital signature
