Only "kmalloc-128" but any other such caches will be on slab. The simplest option would be to make sure that off slab isn't allowedįor caches of KMALLOC_MIN_SIZE or smaller, with the drawback that not The kmalloc_caches we are trying to create. Slab_early_init is already 0), _kmem_cache_create() requires anĪllocation of 32 bytes (freelist_size) which has index 7, hence exactly When trying to create "kmalloc-128" (via create_kmalloc_caches(), Goes off-slab when KMALLOC_MIN_SIZE = 128. Following this, the minimum kmalloc allocation Slab_early_init is cleared after "kmem_cache" and "kmalloc-node" cachesĪre successfully created. > However, the failure in line 2283 shows that the OFF SLAB flag was > mistakenly set anyways!!!! Something must havee cleared slab_early_init? > But the slab_early_init is set which should prevent the use of an off slab > management infrastructure in kmem_cache_create(). > After that commit kmem_cache_create would try to allocate an off slab > management structure which is not available during early boot. This means that the first kmalloc cache to be created, "kmalloc-128", is > The commit allows off slab management for PAGE_SIZE > 5 that is 128. > I'll do some more investigation tomorrow. Another option may be to change > kmalloc_size and kmalloc_index to cope with KMALLOC_MIN_SIZE of 128. But the kmalloc_caches has never > been populated, hence the BUG_ON. The size_index array is overridden so > that 0.15 are 7 and 16.23 are 8. > If I revert commit 8fc9cf420b36 ("slab: make more slab management > structure off the slab") it works but I still need to figure out how > slab indices are calculated. Kmalloc_index(152) returns 8 (INDEX_NODE=8) since KMALLOC_MIN_SIZE=128Īnd the "kmalloc-node" cache size is 256. I'm not entirely sure that gets used (or even created). Yes that would mean use the 196 sized kmalloc array which is not a > power of two slab. > On Tue,, Catalin Marinas wrote: > (cc'ing Jonsoo and Christoph summary: slab failure with L1_CACHE_BYTES > of 128 and sizeof(kmem_cache_node) of 152) > Hmmm. On Tue, at 05:33:25PM -0600, Christoph Lameter wrote: X-Spam-Checker-Version: SpamAssassin 3.3.1 () on Ĭatalin Marinas Nov. UNPARSEABLE_RELAY autolearn=unavailable version=3.3.1 X-Mailing-List: No, score=-6.9 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_HI, Geert Uytterhoeven Re: arm64: Increase the max granular sizeĬontent-Type: text/plain charset=us-ascii Received: from (unknown )īy (Postfix) with ESMTP id C21CA4A Received: from ( )īy (Postfix) with ESMTP id B2308206B7 Received: from (localhost )īy (Postfix) with ESMTP id E12FD206D4 a/arch/arm64/boot/dts/qcom/sdm845.X-Original-To: from ( )īy (Postfix) with ESMTP id ED274BEEA4 While touching all reg properties, addresses are padded to 8 digits. Size cells to make it possible to describe this in "ranges" and The busses on SDM845 provides 36 address bits, extend the address and Arm64: dts: qcom: sdm845: Increase address and size cells for soc - Bjorn Andersson All of help / color / mirror / Atom feed From: Bjorn Andersson Ĭc: arm64: dts: qcom: sdm845: Increase address and size cells for socĭate: Wed, 20:29:39 -0800
0 kommentar(er)
