April 2025 - Linux-stable-mirror

[PATCH] bus: mhi: ep: Update read pointer only after buffer is written

by Sumit Kumar

Inside mhi_ep_ring_add_element, the read pointer (rd_offset) is updated before the buffer is written, potentially causing race conditions where the host sees an updated read pointer before the buffer is actually written. Updating rd_offset prematurely can lead to the host accessing an uninitialized or incomplete element, resulting in data corruption. Invoke the buffer write before updating rd_offset to ensure the element is fully written before signaling its availability. Fixes: bbdcba57a1a2 ("bus: mhi: ep: Add support for ring management") cc: stable(a)vger.kernel.org Co-developed-by: Youssef Samir <quic_yabdulra(a)quicinc.com> Signed-off-by: Youssef Samir <quic_yabdulra(a)quicinc.com> Signed-off-by: Sumit Kumar <quic_sumk(a)quicinc.com> --- --- drivers/bus/mhi/ep/ring.c | 16 ++++++++++------ 1 file changed, 10 insertions(+), 6 deletions(-) diff --git a/drivers/bus/mhi/ep/ring.c b/drivers/bus/mhi/ep/ring.c index aeb53b2c34a8cd859393529d0c8860462bc687ed..26357ee68dee984d70ae5bf39f8f09f2cbcafe30 100644 --- a/drivers/bus/mhi/ep/ring.c +++ b/drivers/bus/mhi/ep/ring.c @@ -131,19 +131,23 @@ int mhi_ep_ring_add_element(struct mhi_ep_ring *ring, struct mhi_ring_element *e } old_offset = ring->rd_offset; - mhi_ep_ring_inc_index(ring); dev_dbg(dev, "Adding an element to ring at offset (%zu)\n", ring->rd_offset); + buf_info.host_addr = ring->rbase + (old_offset * sizeof(*el)); + buf_info.dev_addr = el; + buf_info.size = sizeof(*el); + + ret = mhi_cntrl->write_sync(mhi_cntrl, &buf_info); + if (ret) + return ret; + + mhi_ep_ring_inc_index(ring); /* Update rp in ring context */ rp = cpu_to_le64(ring->rd_offset * sizeof(*el) + ring->rbase); memcpy_toio((void __iomem *) &ring->ring_ctx->generic.rp, &rp, sizeof(u64)); - buf_info.host_addr = ring->rbase + (old_offset * sizeof(*el)); - buf_info.dev_addr = el; - buf_info.size = sizeof(*el); - - return mhi_cntrl->write_sync(mhi_cntrl, &buf_info); + return ret; } void mhi_ep_ring_init(struct mhi_ep_ring *ring, enum mhi_ep_ring_type type, u32 id) --- base-commit: 1e26c5e28ca5821a824e90dd359556f5e9e7b89f change-id: 20250328-rp_fix-d7ebc18bc3be Best regards, -- Sumit Kumar <quic_sumk(a)quicinc.com>

7 months, 2 weeks

4
4
0 0

[PATCH 1/2] PCI/MSI: Add MSIX option to write to ENTRY_DATA before any reads

by dullfire＠yahoo.com

From: Jonathan Currier <dullfire(a)yahoo.com> Commit 7d5ec3d36123 ("PCI/MSI: Mask all unused MSI-X entries") introduces a readl() from ENTRY_VECTOR_CTRL before the writel() to ENTRY_DATA. This is correct, however some hardware, like the Sun Neptune chips, the niu module, will cause an error and/or fatal trap if any MSIX table entry is read before the corresponding ENTRY_DATA field is written to. This patch adds an optional early writel() in msix_prepare_msi_desc(). Cc: stable(a)vger.kernel.org Signed-off-by: Jonathan Currier <dullfire(a)yahoo.com> --- drivers/pci/msi/msi.c | 2 ++ include/linux/pci.h | 2 ++ 2 files changed, 4 insertions(+) diff --git a/drivers/pci/msi/msi.c b/drivers/pci/msi/msi.c index 3a45879d85db..50d87fb5e37f 100644 --- a/drivers/pci/msi/msi.c +++ b/drivers/pci/msi/msi.c @@ -611,6 +611,8 @@ void msix_prepare_msi_desc(struct pci_dev *dev, struct msi_desc *desc) if (desc->pci.msi_attrib.can_mask) { void __iomem *addr = pci_msix_desc_addr(desc); + if (dev->dev_flags & PCI_DEV_FLAGS_MSIX_TOUCH_ENTRY_DATA_FIRST) + writel(0, addr + PCI_MSIX_ENTRY_DATA); desc->pci.msix_ctrl = readl(addr + PCI_MSIX_ENTRY_VECTOR_CTRL); } } diff --git a/include/linux/pci.h b/include/linux/pci.h index 37d97bef060f..b8b95b58d522 100644 --- a/include/linux/pci.h +++ b/include/linux/pci.h @@ -245,6 +245,8 @@ enum pci_dev_flags { PCI_DEV_FLAGS_NO_RELAXED_ORDERING = (__force pci_dev_flags_t) (1 << 11), /* Device does honor MSI masking despite saying otherwise */ PCI_DEV_FLAGS_HAS_MSI_MASKING = (__force pci_dev_flags_t) (1 << 12), + /* Device requires write to PCI_MSIX_ENTRY_DATA before any MSIX reads */ + PCI_DEV_FLAGS_MSIX_TOUCH_ENTRY_DATA_FIRST = (__force pci_dev_flags_t) (1 << 13), }; enum pci_irq_reroute_variant { -- 2.45.2

7 months, 2 weeks

5
7
0 0

[PATCH 2/2] net/niu: niu requires MSIX ENTRY_DATA fields touch before entry reads

by dullfire＠yahoo.com

From: Jonathan Currier <dullfire(a)yahoo.com> Fix niu_try_msix() to not cause a fatal trap on sparc systems. Set PCI_DEV_FLAGS_MSIX_TOUCH_ENTRY_DATA_FIRST on the struct pci_dev to work around a bug in the hardware or firmware. For each vector entry in the msix table, niu chips will cause a fatal trap if any registers in that entry are read before that entries' ENTRY_DATA register is written to. Testing indicates writes to other registers are not sufficient to prevent the fatal trap, however the value does not appear to matter. This only needs to happen once after power up, so simply rebooting into a kernel lacking this fix will NOT cause the trap. NON-RESUMABLE ERROR: Reporting on cpu 64 NON-RESUMABLE ERROR: TPC [0x00000000005f6900] <msix_prepare_msi_desc+0x90/0xa0> NON-RESUMABLE ERROR: RAW [4010000000000016:00000e37f93e32ff:0000000202000080:ffffffffffffffff NON-RESUMABLE ERROR: 0000000800000000:0000000000000000:0000000000000000:0000000000000000] NON-RESUMABLE ERROR: handle [0x4010000000000016] stick [0x00000e37f93e32ff] NON-RESUMABLE ERROR: type [precise nonresumable] NON-RESUMABLE ERROR: attrs [0x02000080] < ASI sp-faulted priv > NON-RESUMABLE ERROR: raddr [0xffffffffffffffff] NON-RESUMABLE ERROR: insn effective address [0x000000c50020000c] NON-RESUMABLE ERROR: size [0x8] NON-RESUMABLE ERROR: asi [0x00] CPU: 64 UID: 0 PID: 745 Comm: kworker/64:1 Not tainted 6.11.5 #63 Workqueue: events work_for_cpu_fn TSTATE: 0000000011001602 TPC: 00000000005f6900 TNPC: 00000000005f6904 Y: 00000000 Not tainted TPC: <msix_prepare_msi_desc+0x90/0xa0> g0: 00000000000002e9 g1: 000000000000000c g2: 000000c50020000c g3: 0000000000000100 g4: ffff8000470307c0 g5: ffff800fec5be000 g6: ffff800047a08000 g7: 0000000000000000 o0: ffff800014feb000 o1: ffff800047a0b620 o2: 0000000000000011 o3: ffff800047a0b620 o4: 0000000000000080 o5: 0000000000000011 sp: ffff800047a0ad51 ret_pc: 00000000005f7128 RPC: <__pci_enable_msix_range+0x3cc/0x460> l0: 000000000000000d l1: 000000000000c01f l2: ffff800014feb0a8 l3: 0000000000000020 l4: 000000000000c000 l5: 0000000000000001 l6: 0000000020000000 l7: ffff800047a0b734 i0: ffff800014feb000 i1: ffff800047a0b730 i2: 0000000000000001 i3: 000000000000000d i4: 0000000000000000 i5: 0000000000000000 i6: ffff800047a0ae81 i7: 00000000101888b0 I7: <niu_try_msix.constprop.0+0xc0/0x130 [niu]> Call Trace: [<00000000101888b0>] niu_try_msix.constprop.0+0xc0/0x130 [niu] [<000000001018f840>] niu_get_invariants+0x183c/0x207c [niu] [<00000000101902fc>] niu_pci_init_one+0x27c/0x2fc [niu] [<00000000005ef3e4>] local_pci_probe+0x28/0x74 [<0000000000469240>] work_for_cpu_fn+0x8/0x1c [<000000000046b008>] process_scheduled_works+0x144/0x210 [<000000000046b518>] worker_thread+0x13c/0x1c0 [<00000000004710e0>] kthread+0xb8/0xc8 [<00000000004060c8>] ret_from_fork+0x1c/0x2c [<0000000000000000>] 0x0 Kernel panic - not syncing: Non-resumable error. Fixes: 7d5ec3d36123 ("PCI/MSI: Mask all unused MSI-X entries") Cc: stable(a)vger.kernel.org Signed-off-by: Jonathan Currier <dullfire(a)yahoo.com> --- drivers/net/ethernet/sun/niu.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/net/ethernet/sun/niu.c b/drivers/net/ethernet/sun/niu.c index 41a27ae58ced..f5449b73b9a7 100644 --- a/drivers/net/ethernet/sun/niu.c +++ b/drivers/net/ethernet/sun/niu.c @@ -9058,6 +9058,8 @@ static void niu_try_msix(struct niu *np, u8 *ldg_num_map) msi_vec[i].entry = i; } + pdev->dev_flags |= PCI_DEV_FLAGS_MSIX_TOUCH_ENTRY_DATA_FIRST; + num_irqs = pci_enable_msix_range(pdev, msi_vec, 1, num_irqs); if (num_irqs < 0) { np->flags &= ~NIU_FLAGS_MSIX; -- 2.45.2

7 months, 2 weeks

2
1
0 0

[PATCH v2] sparc: fix error handling in scan_one_device()

by Ma Ke

Once of_device_register() failed, we should call put_device() to decrement reference count for cleanup. Or it could cause memory leak. So fix this by calling put_device(), then the name can be freed in kobject_cleanup(). Calling path: of_device_register() -> of_device_add() -> device_add(). As comment of device_add() says, 'if device_add() succeeds, you should call device_del() when you want to get rid of it. If device_add() has not succeeded, use only put_device() to drop the reference count'. Found by code review. Cc: stable(a)vger.kernel.org Fixes: cf44bbc26cf1 ("[SPARC]: Beginnings of generic of_device framework.") Signed-off-by: Ma Ke <make24(a)iscas.ac.cn> --- Changes in v2: - retained kfree() manually due to the lack of a release callback function. --- arch/sparc/kernel/of_device_64.c | 1 + 1 file changed, 1 insertion(+) diff --git a/arch/sparc/kernel/of_device_64.c b/arch/sparc/kernel/of_device_64.c index f98c2901f335..f53092b07b9e 100644 --- a/arch/sparc/kernel/of_device_64.c +++ b/arch/sparc/kernel/of_device_64.c @@ -677,6 +677,7 @@ static struct platform_device * __init scan_one_device(struct device_node *dp, if (of_device_register(op)) { printk("%pOF: Could not register of device.\n", dp); + put_device(&op->dev); kfree(op); op = NULL; } -- 2.25.1

7 months, 2 weeks

1
0
0 0

[PATCH] ksmbd: Prevent integer overflow in calculation of deadtime

by Denis Arefev

The user can set any value for 'deadtime'. This affects the arithmetic expression 'req->deadtime * SMB_ECHO_INTERVAL', which is subject to overflow. The added check makes the server behavior more predictable. Found by Linux Verification Center (linuxtesting.org) with SVACE. Fixes: 0626e6641f6b ("cifsd: add server handler for central processing and tranport layers") Cc: stable(a)vger.kernel.org Signed-off-by: Denis Arefev <arefev(a)swemel.ru> --- fs/smb/server/transport_ipc.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/fs/smb/server/transport_ipc.c b/fs/smb/server/transport_ipc.c index 3f185ae60dc5..2a3e2b0ce557 100644 --- a/fs/smb/server/transport_ipc.c +++ b/fs/smb/server/transport_ipc.c @@ -310,7 +310,11 @@ static int ipc_server_config_on_startup(struct ksmbd_startup_request *req) server_conf.signing = req->signing; server_conf.tcp_port = req->tcp_port; server_conf.ipc_timeout = req->ipc_timeout * HZ; - server_conf.deadtime = req->deadtime * SMB_ECHO_INTERVAL; + if (check_mul_overflow(req->deadtime, SMB_ECHO_INTERVAL, + &server_conf.deadtime)) { + ret = -EINVAL; + goto out; + } server_conf.share_fake_fscaps = req->share_fake_fscaps; ksmbd_init_domain(req->sub_auth); @@ -337,6 +341,7 @@ static int ipc_server_config_on_startup(struct ksmbd_startup_request *req) server_conf.bind_interfaces_only = req->bind_interfaces_only; ret |= ksmbd_tcp_set_interfaces(KSMBD_STARTUP_CONFIG_INTERFACES(req), req->ifc_list_sz); +out: if (ret) { pr_err("Server configuration error: %s %s %s\n", req->netbios_name, req->server_string, -- 2.43.0

7 months, 2 weeks

2
1
0 0

+ v2-ocfs2-fix-panic-in-failed-foilio-allocation.patch added to mm-nonmm-unstable branch

by Andrew Morton

The patch titled Subject: ocfs2: fix panic in failed foilio allocation has been added to the -mm mm-nonmm-unstable branch. Its filename is v2-ocfs2-fix-panic-in-failed-foilio-allocation.patch This patch will shortly appear at https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patche… This patch will later appear in the mm-nonmm-unstable branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm Before you just go and hit "reply", please: a) Consider who else should be cc'ed b) Prefer to cc a suitable mailing list as well c) Ideally: find the original patch on the mailing list and do a reply-to-all to that, adding suitable additional cc's *** Remember to use Documentation/process/submit-checklist.rst when testing your code *** The -mm tree is included into linux-next via the mm-everything branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm and is updated there every 2-3 working days ------------------------------------------------------ From: Mark Tinguely <mark.tinguely(a)oracle.com> Subject: ocfs2: fix panic in failed foilio allocation Date: Fri, 11 Apr 2025 11:31:24 -0500 In the page to order 0 folio conversion series, commits 7e119cff9d0a ("ocfs2: convert w_pages to w_folios") and 9a5e08652dc4 ("ocfs2: use an array of folios instead of an array of pages") save -ENOMEM in the folio array upon allocation failure and call the folio array free code. The folio array free code expects either valid folio pointers or NULL. Finding the -ENOMEM will result in a panic. Fix by NULLing the error folio entry. Link: https://lkml.kernel.org/r/c879a52b-835c-4fa0-902b-8b2e9196dcbd@oracle.com Fixes: 7e119cff9d0a ("ocfs2: convert w_pages to w_folios") FIxes: 9a5e08652dc4 ("ocfs2: use an array of folios instead of an array of pages") Signed-off-by: Mark Tinguely <mark.tinguely(a)oracle.com> Reviewed-by: Matthew Wilcox (Oracle) <willy(a)infradead.org> Reviewed-by: Joseph Qi <joseph.qi(a)linux.alibaba.com> Cc: Changwei Ge <gechangwei(a)live.cn> Cc: Joel Becker <jlbec(a)evilplan.org> Cc: Junxiao Bi <junxiao.bi(a)oracle.com> Cc: Mark Fasheh <mark(a)fasheh.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- fs/ocfs2/alloc.c | 1 + fs/ocfs2/aops.c | 1 + 2 files changed, 2 insertions(+) --- a/fs/ocfs2/alloc.c~v2-ocfs2-fix-panic-in-failed-foilio-allocation +++ a/fs/ocfs2/alloc.c @@ -6918,6 +6918,7 @@ static int ocfs2_grab_folios(struct inod if (IS_ERR(folios[numfolios])) { ret = PTR_ERR(folios[numfolios]); mlog_errno(ret); + folios[numfolios] = NULL; goto out; } --- a/fs/ocfs2/aops.c~v2-ocfs2-fix-panic-in-failed-foilio-allocation +++ a/fs/ocfs2/aops.c @@ -1071,6 +1071,7 @@ static int ocfs2_grab_folios_for_write(s if (IS_ERR(wc->w_folios[i])) { ret = PTR_ERR(wc->w_folios[i]); mlog_errno(ret); + wc->w_folios[i] = NULL; goto out; } } _ Patches currently in -mm which might be from mark.tinguely(a)oracle.com are v2-ocfs2-fix-panic-in-failed-foilio-allocation.patch

7 months, 2 weeks

1
0
0 0

[PATCH] f2fs: prevent kernel warning due to negative i_nlink from corrupted image

by Jaegeuk Kim

WARNING: CPU: 1 PID: 9426 at fs/inode.c:417 drop_nlink+0xac/0xd0 home/cc/linux/fs/inode.c:417 Modules linked in: CPU: 1 UID: 0 PID: 9426 Comm: syz-executor568 Not tainted 6.14.0-12627-g94d471a4f428 #2 PREEMPT(full) Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.13.0-1ubuntu1.1 04/01/2014 RIP: 0010:drop_nlink+0xac/0xd0 home/cc/linux/fs/inode.c:417 Code: 48 8b 5d 28 be 08 00 00 00 48 8d bb 70 07 00 00 e8 f9 67 e6 ff f0 48 ff 83 70 07 00 00 5b 5d e9 9a 12 82 ff e8 95 12 82 ff 90 <0f> 0b 90 c7 45 48 ff ff ff ff 5b 5d e9 83 12 82 ff e8 fe 5f e6 ff RSP: 0018:ffffc900026b7c28 EFLAGS: 00010293 RAX: 0000000000000000 RBX: 0000000000000000 RCX: ffffffff8239710f RDX: ffff888041345a00 RSI: ffffffff8239717b RDI: 0000000000000005 RBP: ffff888054509ad0 R08: 0000000000000005 R09: 0000000000000000 R10: 0000000000000000 R11: ffffffff9ab36f08 R12: ffff88804bb40000 R13: ffff8880545091e0 R14: 0000000000008000 R15: ffff8880545091e0 FS: 000055555d0c5880(0000) GS:ffff8880eb3e3000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007f915c55b178 CR3: 0000000050d20000 CR4: 0000000000352ef0 Call Trace: <task> f2fs_i_links_write home/cc/linux/fs/f2fs/f2fs.h:3194 [inline] f2fs_drop_nlink+0xd1/0x3c0 home/cc/linux/fs/f2fs/dir.c:845 f2fs_delete_entry+0x542/0x1450 home/cc/linux/fs/f2fs/dir.c:909 f2fs_unlink+0x45c/0x890 home/cc/linux/fs/f2fs/namei.c:581 vfs_unlink+0x2fb/0x9b0 home/cc/linux/fs/namei.c:4544 do_unlinkat+0x4c5/0x6a0 home/cc/linux/fs/namei.c:4608 __do_sys_unlink home/cc/linux/fs/namei.c:4654 [inline] __se_sys_unlink home/cc/linux/fs/namei.c:4652 [inline] __x64_sys_unlink+0xc5/0x110 home/cc/linux/fs/namei.c:4652 do_syscall_x64 home/cc/linux/arch/x86/entry/syscall_64.c:63 [inline] do_syscall_64+0xc7/0x250 home/cc/linux/arch/x86/entry/syscall_64.c:94 entry_SYSCALL_64_after_hwframe+0x77/0x7f RIP: 0033:0x7fb3d092324b Code: 73 01 c3 48 c7 c1 c0 ff ff ff f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa b8 57 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 c0 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007ffdc232d938 EFLAGS: 00000206 ORIG_RAX: 0000000000000057 RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fb3d092324b RDX: 00007ffdc232d960 RSI: 00007ffdc232d960 RDI: 00007ffdc232d9f0 RBP: 00007ffdc232d9f0 R08: 0000000000000001 R09: 00007ffdc232d7c0 R10: 00000000fffffffd R11: 0000000000000206 R12: 00007ffdc232eaf0 R13: 000055555d0cebb0 R14: 00007ffdc232d958 R15: 0000000000000001 </task> Cc: stable(a)vger.kernel.org Signed-off-by: Jaegeuk Kim <jaegeuk(a)kernel.org> --- fs/f2fs/namei.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/fs/f2fs/namei.c b/fs/f2fs/namei.c index 8f8b9b843bdf..f17cb2489a73 100644 --- a/fs/f2fs/namei.c +++ b/fs/f2fs/namei.c @@ -569,6 +569,15 @@ static int f2fs_unlink(struct inode *dir, struct dentry *dentry) goto fail; } + if (unlikely(inode->i_nlink == 0)) { + f2fs_warn(F2FS_I_SB(inode), "%s: inode (ino=%lx) has zero i_nlink", + __func__, inode->i_ino); + err = -EFSCORRUPTED; + set_sbi_flag(F2FS_I_SB(inode), SBI_NEED_FSCK); + f2fs_put_page(page, 0); + goto fail; + } + f2fs_balance_fs(sbi, true); f2fs_lock_op(sbi); -- 2.49.0.604.gff1f9ca942-goog

7 months, 2 weeks

3
2
0 0

[PATCH RESEND] sparc: fix error handling in scan_one_device()

by Ma Ke

Once of_device_register() failed, we should call put_device() to decrement reference count for cleanup. Or it could cause memory leak. So fix this by calling put_device(), then the name can be freed in kobject_cleanup(). As comment of device_add() says, 'if device_add() succeeds, you should call device_del() when you want to get rid of it. If device_add() has not succeeded, use only put_device() to drop the reference count'. Found by code review. Cc: stable(a)vger.kernel.org Fixes: cf44bbc26cf1 ("[SPARC]: Beginnings of generic of_device framework.") Signed-off-by: Ma Ke <make24(a)iscas.ac.cn> --- arch/sparc/kernel/of_device_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/sparc/kernel/of_device_64.c b/arch/sparc/kernel/of_device_64.c index f98c2901f335..4272746d7166 100644 --- a/arch/sparc/kernel/of_device_64.c +++ b/arch/sparc/kernel/of_device_64.c @@ -677,7 +677,7 @@ static struct platform_device * __init scan_one_device(struct device_node *dp, if (of_device_register(op)) { printk("%pOF: Could not register of device.\n", dp); - kfree(op); + put_device(&op->dev); op = NULL; } -- 2.25.1

7 months, 2 weeks

2
1
0 0

Re: Patch "x86/mm/ident_map: Fix theoretical virtual address overflow to zero" has been added to the 6.14-stable tree

by Kirill A. Shutemov

On Mon, Apr 14, 2025 at 06:37:27AM -0400, Sasha Levin wrote: > This is a note to let you know that I've just added the patch titled > > x86/mm/ident_map: Fix theoretical virtual address overflow to zero > > to the 6.14-stable tree which can be found at: > http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum… > > The filename of the patch is: > x86-mm-ident_map-fix-theoretical-virtual-address-ove.patch > and it can be found in the queue-6.14 subdirectory. > > If you, or anyone else, feels it should not be added to the stable tree, > please let <stable(a)vger.kernel.org> know about it. It was explicitly called out that no backporting needed: > [ Backporter's note: there's no need to backport this patch. ] -- Kiryl Shutsemau / Kirill A. Shutemov

7 months, 2 weeks

2
3
0 0

[PATCH] EDAC/amd64: Fix size calculation for Non-Power-of-Two DIMMs

by Avadhut Naik

Each Chip-Select (CS) of a Unified Memory Controller (UMC) on AMD EPYC SOCs has an Address Mask and a Secondary Address Mask register associated with it. The amd64_edac module logs DIMM sizes on a per-UMC per-CS granularity during init using these two registers. Currently, the module primarily considers only the Address Mask register for computing DIMM sizes. The Secondary Address Mask register is only considered for odd CS. Additionally, if it has been considered, the Address Mask register is ignored altogether for that CS. For power-of-two DIMMs, this is not an issue since only the Address Mask register is used. For non-power-of-two DIMMs, however, the Secondary Address Mask register is used in conjunction with the Address Mask register. However, since the module only considers either of the two registers for a CS, the size computed by the module is incorrect. The Secondary Address Mask register is not considered for even CS, and the Address Mask register is not considered for odd CS. Furthermore, also rename some variables for greater clarity. Fixes: 81f5090db843 ("EDAC/amd64: Support asymmetric dual-rank DIMMs") Signed-off-by: Avadhut Naik <avadhut.naik(a)amd.com> Cc: stable(a)vger.kernel.org --- drivers/edac/amd64_edac.c | 56 ++++++++++++++++++++++++--------------- 1 file changed, 35 insertions(+), 21 deletions(-) diff --git a/drivers/edac/amd64_edac.c b/drivers/edac/amd64_edac.c index 90f0eb7cc5b9..16117fda727f 100644 --- a/drivers/edac/amd64_edac.c +++ b/drivers/edac/amd64_edac.c @@ -1209,7 +1209,9 @@ static int umc_get_cs_mode(int dimm, u8 ctrl, struct amd64_pvt *pvt) if (csrow_enabled(2 * dimm + 1, ctrl, pvt)) cs_mode |= CS_ODD_PRIMARY; - /* Asymmetric dual-rank DIMM support. */ + if (csrow_sec_enabled(2 * dimm, ctrl, pvt)) + cs_mode |= CS_EVEN_SECONDARY; + if (csrow_sec_enabled(2 * dimm + 1, ctrl, pvt)) cs_mode |= CS_ODD_SECONDARY; @@ -1230,12 +1232,10 @@ static int umc_get_cs_mode(int dimm, u8 ctrl, struct amd64_pvt *pvt) return cs_mode; } -static int __addr_mask_to_cs_size(u32 addr_mask_orig, unsigned int cs_mode, - int csrow_nr, int dimm) +static int calculate_cs_size(u32 mask, unsigned int cs_mode) { - u32 msb, weight, num_zero_bits; - u32 addr_mask_deinterleaved; - int size = 0; + int msb, weight, num_zero_bits; + u32 deinterleaved_mask = 0; /* * The number of zero bits in the mask is equal to the number of bits @@ -1248,19 +1248,32 @@ static int __addr_mask_to_cs_size(u32 addr_mask_orig, unsigned int cs_mode, * without swapping with the most significant bit. This can be handled * by keeping the MSB where it is and ignoring the single zero bit. */ - msb = fls(addr_mask_orig) - 1; - weight = hweight_long(addr_mask_orig); + msb = fls(mask) - 1; + weight = hweight_long(mask); num_zero_bits = msb - weight - !!(cs_mode & CS_3R_INTERLEAVE); /* Take the number of zero bits off from the top of the mask. */ - addr_mask_deinterleaved = GENMASK_ULL(msb - num_zero_bits, 1); + deinterleaved_mask = GENMASK(msb - num_zero_bits, 1); + edac_dbg(1, " Deinterleaved AddrMask: 0x%x\n", deinterleaved_mask); + + return (deinterleaved_mask >> 2) + 1; +} + +static int __addr_mask_to_cs_size(u32 addr_mask, u32 addr_mask_sec, + unsigned int cs_mode, int csrow_nr, int dimm) +{ + int size = 0; edac_dbg(1, "CS%d DIMM%d AddrMasks:\n", csrow_nr, dimm); - edac_dbg(1, " Original AddrMask: 0x%x\n", addr_mask_orig); - edac_dbg(1, " Deinterleaved AddrMask: 0x%x\n", addr_mask_deinterleaved); + edac_dbg(1, " Primary AddrMask: 0x%x\n", addr_mask); /* Register [31:1] = Address [39:9]. Size is in kBs here. */ - size = (addr_mask_deinterleaved >> 2) + 1; + size = calculate_cs_size(addr_mask, cs_mode); + + if (addr_mask_sec) { + edac_dbg(1, " Secondary AddrMask: 0x%x\n", addr_mask); + size += calculate_cs_size(addr_mask_sec, cs_mode); + } /* Return size in MBs. */ return size >> 10; @@ -1270,7 +1283,7 @@ static int umc_addr_mask_to_cs_size(struct amd64_pvt *pvt, u8 umc, unsigned int cs_mode, int csrow_nr) { int cs_mask_nr = csrow_nr; - u32 addr_mask_orig; + u32 addr_mask = 0, addr_mask_sec = 0; int dimm, size = 0; /* No Chip Selects are enabled. */ @@ -1308,13 +1321,13 @@ static int umc_addr_mask_to_cs_size(struct amd64_pvt *pvt, u8 umc, if (!pvt->flags.zn_regs_v2) cs_mask_nr >>= 1; - /* Asymmetric dual-rank DIMM support. */ - if ((csrow_nr & 1) && (cs_mode & CS_ODD_SECONDARY)) - addr_mask_orig = pvt->csels[umc].csmasks_sec[cs_mask_nr]; - else - addr_mask_orig = pvt->csels[umc].csmasks[cs_mask_nr]; + if (cs_mode & CS_EVEN_PRIMARY || cs_mode & CS_ODD_PRIMARY) + addr_mask = pvt->csels[umc].csmasks[cs_mask_nr]; + + if (cs_mode & CS_EVEN_SECONDARY || cs_mode & CS_ODD_SECONDARY) + addr_mask_sec = pvt->csels[umc].csmasks_sec[cs_mask_nr]; - return __addr_mask_to_cs_size(addr_mask_orig, cs_mode, csrow_nr, dimm); + return __addr_mask_to_cs_size(addr_mask, addr_mask_sec, cs_mode, csrow_nr, dimm); } static void umc_debug_display_dimm_sizes(struct amd64_pvt *pvt, u8 ctrl) @@ -3512,9 +3525,10 @@ static void gpu_get_err_info(struct mce *m, struct err_info *err) static int gpu_addr_mask_to_cs_size(struct amd64_pvt *pvt, u8 umc, unsigned int cs_mode, int csrow_nr) { - u32 addr_mask_orig = pvt->csels[umc].csmasks[csrow_nr]; + u32 addr_mask = pvt->csels[umc].csmasks[csrow_nr]; + u32 addr_mask_sec = pvt->csels[umc].csmasks_sec[csrow_nr]; - return __addr_mask_to_cs_size(addr_mask_orig, cs_mode, csrow_nr, csrow_nr >> 1); + return __addr_mask_to_cs_size(addr_mask, addr_mask_sec, cs_mode, csrow_nr, csrow_nr >> 1); } static void gpu_debug_display_dimm_sizes(struct amd64_pvt *pvt, u8 ctrl) base-commit: f1861fb575b028e35e6233295441d535f2e3f240 -- 2.43.0

7 months, 2 weeks

3
4
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror April 2025