user/sven/linux.git/drivers/ntb, branch v6.1.162

NTB: epf: Allow arbitrary BAR mapping

2025-12-06T21:12:28Z

[ Upstream commit 5ad865862a0fd349163243e1834ed98ba9b81905 ] The NTB epf host driver assumes the BAR number associated with a memory window is just incremented from the BAR number associated with MW1. This seems to have been enough so far but this is not really how the endpoint side work and the two could easily become mis-aligned. ntb_epf_mw_to_bar() even assumes that the BAR number is the memory window index + 2, which means the function only returns a proper result if BAR_2 is associated with MW1. Instead, fully describe and allow arbitrary NTB BAR mapping. Signed-off-by: Jerome Brunet Signed-off-by: Jon Mason Signed-off-by: Sasha Levin

ntb_hw_amd: Add NTB PCI ID for new gen CPU

2025-05-02T05:47:04Z

[ Upstream commit bf8a7ce7e4c7267a6f5f2b2023cfc459b330b25e ] Add NTB support for new generation of processor. Signed-off-by: Basavaraj Natikar Signed-off-by: Jon Mason Signed-off-by: Sasha Levin

ntb: reduce stack usage in idt_scan_mws

2025-05-02T05:47:04Z

[ Upstream commit aff12700b8dd7422bfe2277696e192af4df9de8f ] idt_scan_mws() puts a large fixed-size array on the stack and copies it into a smaller dynamically allocated array at the end. On 32-bit targets, the fixed size can easily exceed the warning limit for possible stack overflow: drivers/ntb/hw/idt/ntb_hw_idt.c:1041:27: error: stack frame size (1032) exceeds limit (1024) in 'idt_scan_mws' [-Werror,-Wframe-larger-than] Change it to instead just always use dynamic allocation for the array from the start. It's too big for the stack, but not actually all that much for a permanent allocation. Reported-by: kernel test robot Closes: https://lore.kernel.org/all/202205111109.PiKTruEj-lkp@intel.com/ Signed-off-by: Arnd Bergmann Reviewed-by: Dave Jiang Reviewed-by: Damien Le Moal Signed-off-by: Jon Mason Signed-off-by: Sasha Levin

ntb: use 64-bit arithmetic for the MSI doorbell mask

2025-04-25T08:43:45Z

commit fd5625fc86922f36bedee5846fefd647b7e72751 upstream. msi_db_mask is of type 'u64', still the standard 'int' arithmetic is performed to compute its value. While most of the ntb_hw drivers actually don't utilize the higher 32 bits of the doorbell mask now, this may be the case for Switchtec - see switchtec_ntb_init_db(). Found by Linux Verification Center (linuxtesting.org) with SVACE static analysis tool. Fixes: 2b0569b3b7e6 ("NTB: Add MSI interrupt support to ntb_transport") Cc: stable@vger.kernel.org Signed-off-by: Fedor Pchelkin Reviewed-by: Dave Jiang Signed-off-by: Jon Mason Signed-off-by: Greg Kroah-Hartman

ntb_perf: Delete duplicate dmaengine_unmap_put() call in perf_copy_chunk()

2025-04-10T12:33:42Z

commit 4279e72cab31dd3eb8c89591eb9d2affa90ab6aa upstream. The function call “dmaengine_unmap_put(unmap)” was used in an if branch. The same call was immediately triggered by a subsequent goto statement. Thus avoid such a call repetition. This issue was detected by using the Coccinelle software. Fixes: 5648e56d03fa ("NTB: ntb_perf: Add full multi-port NTB API support") Cc: stable@vger.kernel.org Signed-off-by: Markus Elfring Signed-off-by: Jon Mason Signed-off-by: Greg Kroah-Hartman

ntb: intel: Fix using link status DB's

2025-04-10T12:33:40Z

[ Upstream commit 8144e9c8f30fb23bb736a5d24d5c9d46965563c4 ] Make sure we are not using DB's which were remapped for link status. Fixes: f6e51c354b60 ("ntb: intel: split out the gen3 code") Signed-off-by: Nikita Shubin Reviewed-by: Dave Jiang Signed-off-by: Jon Mason Signed-off-by: Sasha Levin

ntb_hw_switchtec: Fix shift-out-of-bounds in switchtec_ntb_mw_set_trans

2025-04-10T12:33:40Z

[ Upstream commit de203da734fae00e75be50220ba5391e7beecdf9 ] There is a kernel API ntb_mw_clear_trans() would pass 0 to both addr and size. This would make xlate_pos negative. [ 23.734156] switchtec switchtec0: MW 0: part 0 addr 0x0000000000000000 size 0x0000000000000000 [ 23.734158] ================================================================================ [ 23.734172] UBSAN: shift-out-of-bounds in drivers/ntb/hw/mscc/ntb_hw_switchtec.c:293:7 [ 23.734418] shift exponent -1 is negative Ensuring xlate_pos is a positive or zero before BIT. Fixes: 1e2fd202f859 ("ntb_hw_switchtec: Check for alignment of the buffer in mw_set_trans()") Signed-off-by: Yajun Deng Reviewed-by: Logan Gunthorpe Signed-off-by: Jon Mason Signed-off-by: Sasha Levin

ntb: ntb_hw_switchtec: Fix use after free vulnerability in switchtec_ntb_remove due to race condition

2024-10-17T13:22:15Z

[ Upstream commit e51aded92d42784313ba16c12f4f88cc4f973bbb ] In the switchtec_ntb_add function, it can call switchtec_ntb_init_sndev function, then &sndev->check_link_status_work is bound with check_link_status_work. switchtec_ntb_link_notification may be called to start the work. If we remove the module which will call switchtec_ntb_remove to make cleanup, it will free sndev through kfree(sndev), while the work mentioned above will be used. The sequence of operations that may lead to a UAF bug is as follows: CPU0 CPU1 | check_link_status_work switchtec_ntb_remove | kfree(sndev); | | if (sndev->link_force_down) | // use sndev Fix it by ensuring that the work is canceled before proceeding with the cleanup in switchtec_ntb_remove. Signed-off-by: Kaixin Wang Reviewed-by: Logan Gunthorpe Signed-off-by: Jon Mason Signed-off-by: Sasha Levin

ntb: Force physically contiguous allocation of rx ring buffers

2024-10-17T13:21:08Z

[ Upstream commit 061a785a114f159e990ea8ed8d1b7dca4b41120f ] Physical addresses under IOVA on x86 platform are mapped contiguously as a side effect before the patch that removed CONFIG_DMA_REMAP. The NTB rx buffer ring is a single chunk DMA buffer that is allocated against the NTB PCI device. If the receive side is using a DMA device, then the buffers are remapped against the DMA device before being submitted via the dmaengine API. This scheme becomes a problem when the physical memory is discontiguous. When dma_map_page() is called on the kernel virtual address from the dma_alloc_coherent() call, the new IOVA mapping no longer points to all the physical memory allocated due to being discontiguous. Change dma_alloc_coherent() to dma_alloc_attrs() in order to force DMA_ATTR_FORCE_CONTIGUOUS attribute. This is the best fix for the circumstance. A potential future solution may be having the DMA mapping API providing a way to alias an existing IOVA mapping to a new device perhaps. This fix is not to fix the patch pointed to by the fixes tag, but to fix the issue arised in the ntb_transport driver on x86 platforms after the said patch is applied. Reported-by: Jerry Dai Fixes: f5ff79fddf0e ("dma-mapping: remove CONFIG_DMA_REMAP") Tested-by: Jerry Dai Signed-off-by: Dave Jiang Signed-off-by: Jon Mason Signed-off-by: Sasha Levin

ntb_perf: Fix printk format

2024-10-17T13:21:08Z

[ Upstream commit 1501ae7479c8d0f66efdbfdc9ae8d6136cefbd37 ] The correct printk format is %pa or %pap, but not %pa[p]. Fixes: 99a06056124d ("NTB: ntb_perf: Fix address err in perf_copy_chunk") Signed-off-by: Max Hawking Signed-off-by: Jon Mason Signed-off-by: Sasha Levin