From 93de2a6a4b91b72607136dd656edf03fb399d27f Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Wed, 23 Sep 2026 09:37:21 -0700 Subject: KVM: SEV: Do cache maintenance on the source VM during intra-host migration Manually perform cache maintenance on the source VM during intra-host migration to ensure no stale data is left in CPU caches after the VM is destroyed. Because the source VM is "converted" to a non-SEV VM, KVM's memory reclaim flows won't trigger cache maintenance, e.g. when all guest memory is reclaimed in response to detaching from the mmu_notifier. Note, relying on the destination VM to do cache maintenance isn't an option as KVM doesn't require identical guest memory configurations, i.e. the source VM may have access to memory that the destination VM does not. Enforcing equivalent memory configurations is infeasible, as it would require a *deep* comparison of memslots, e.g. to verify that not only are the memslot identical, but what the memslots point at is also identical. Fixes: b56639318bb2 ("KVM: SEV: Add support for SEV intra host migration") Cc: stable@vger.kernel.org Reported-by: Stefan Teodorescu Signed-off-by: Sean Christopherson Message-ID: <20260923163721.1584779-3-seanjc@google.com> Signed-off-by: Paolo Bonzini --- fs/smb/client/smb2file.c | 470 +++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 470 insertions(+) create mode 100644 fs/smb/client/smb2file.c (limited to 'fs/smb/client/smb2file.c') diff --git a/fs/smb/client/smb2file.c b/fs/smb/client/smb2file.c new file mode 100644 index 000000000..fb2fccbe8 --- /dev/null +++ b/fs/smb/client/smb2file.c @@ -0,0 +1,470 @@ +// SPDX-License-Identifier: LGPL-2.1 +/* + * + * Copyright (C) International Business Machines Corp., 2002, 2011 + * Author(s): Steve French (sfrench@us.ibm.com), + * Pavel Shilovsky ((pshilovsky@samba.org) 2012 + * + */ +#include +#include +#include +#include +#include +#include +#include "cifsfs.h" +#include "cifsglob.h" +#include "cifsproto.h" +#include "cifs_debug.h" +#include "cifs_fs_sb.h" +#include "cifs_unicode.h" +#include "fscache.h" +#include "smb2proto.h" +#include "../common/smb2status.h" +#include "../common/smbfsctl.h" + +static struct smb2_symlink_err_rsp *symlink_data(const struct kvec *iov) +{ + struct smb2_err_rsp *err = iov->iov_base; + struct smb2_symlink_err_rsp *sym = ERR_PTR(-EINVAL); + u8 *end = (u8 *)err + iov->iov_len; + u32 len; + + /* + * Per [MS-SMB2] section 2.2.2, a STATUS_STOPPED_ON_SYMLINK response has to + * carry a Symbolic Link Error Response, so ByteCount cannot be zero. Some + * servers (e.g. the macOS built-in SMB server) violate this and return an + * empty error response, with both ErrorContextCount and ByteCount set to + * zero, i.e. without the symlink target. Detect this and return -ENODATA + * so that callers can tell "server did not send the target" apart from a + * malformed response, and retrieve the target with FSCTL_GET_REPARSE_POINT + * instead. + */ + if (!err->ErrorContextCount && !le32_to_cpu(err->ByteCount)) + return ERR_PTR(-ENODATA); + + if (err->ErrorContextCount) { + struct smb2_error_context_rsp *p; + + len = (u32)err->ErrorContextCount * (offsetof(struct smb2_error_context_rsp, + ErrorContextData) + + sizeof(struct smb2_symlink_err_rsp)); + if (le32_to_cpu(err->ByteCount) < len || iov->iov_len < len + sizeof(*err) + 1) + return ERR_PTR(-EINVAL); + + p = (struct smb2_error_context_rsp *)err->ErrorData; + while ((u8 *)p + sizeof(*p) <= end) { + if (le32_to_cpu(p->ErrorId) == SMB2_ERROR_ID_DEFAULT) { + sym = (struct smb2_symlink_err_rsp *)p->ErrorContextData; + break; + } + cifs_dbg(FYI, "%s: skipping unhandled error context: 0x%x\n", + __func__, le32_to_cpu(p->ErrorId)); + + len = le32_to_cpu(p->ErrorDataLength); + if (len > end - ((u8 *)p + sizeof(*p))) + return ERR_PTR(-EINVAL); + len = ALIGN(len, 8); + if (len > end - ((u8 *)p + sizeof(*p))) + return ERR_PTR(-EINVAL); + + p = (struct smb2_error_context_rsp *)(p->ErrorContextData + len); + } + } else if (le32_to_cpu(err->ByteCount) >= sizeof(*sym) && + iov->iov_len >= SMB2_SYMLINK_STRUCT_SIZE) { + sym = (struct smb2_symlink_err_rsp *)err->ErrorData; + } + + if (!IS_ERR(sym) && + ((u8 *)sym + sizeof(*sym) > end || + le32_to_cpu(sym->SymLinkErrorTag) != SYMLINK_ERROR_TAG || + le32_to_cpu(sym->ReparseTag) != IO_REPARSE_TAG_SYMLINK)) + sym = ERR_PTR(-EINVAL); + + return sym; +} + +int smb2_fix_symlink_target_type(char **target, bool directory, struct cifs_sb_info *cifs_sb) +{ + char *buf; + int len; + + /* + * POSIX server does not distinguish between symlinks to file and + * symlink directory. So nothing is needed to fix on the client side. + */ + if (cifs_sb_flags(cifs_sb) & CIFS_MOUNT_POSIX_PATHS) + return 0; + + if (!*target) + return smb_EIO(smb_eio_trace_null_pointers); + + len = strlen(*target); + if (!len) + return smb_EIO1(smb_eio_trace_sym_target_len, len); + + /* + * If this is directory symlink and it does not have trailing slash then + * append it. Trailing slash simulates Windows/SMB behavior which do not + * allow resolving directory symlink to file. + */ + if (directory && (*target)[len-1] != '/') { + buf = krealloc(*target, len+2, GFP_KERNEL); + if (!buf) + return -ENOMEM; + buf[len] = '/'; + buf[len+1] = '\0'; + *target = buf; + len++; + } + + /* + * If this is a file (non-directory) symlink and it points to path name + * with trailing slash then this is an invalid symlink because file name + * cannot contain slash character. File name with slash is invalid on + * both Windows and Linux systems. So return an error for such symlink. + */ + if (!directory && (*target)[len-1] == '/') + return smb_EIO(smb_eio_trace_sym_slash); + + return 0; +} + +int smb2_parse_symlink_response(struct cifs_sb_info *cifs_sb, const struct kvec *iov, + const char *full_path, char **path) +{ + struct smb2_symlink_err_rsp *sym; + unsigned int sub_offs, sub_len; + unsigned int print_offs, print_len; + + if (!cifs_sb || !iov || !iov->iov_base || !iov->iov_len || !path) + return -EINVAL; + + sym = symlink_data(iov); + if (IS_ERR(sym)) + return PTR_ERR(sym); + + sub_len = le16_to_cpu(sym->SubstituteNameLength); + sub_offs = le16_to_cpu(sym->SubstituteNameOffset); + print_len = le16_to_cpu(sym->PrintNameLength); + print_offs = le16_to_cpu(sym->PrintNameOffset); + + if ((char *)sym->PathBuffer + sub_offs + sub_len > + (char *)iov->iov_base + iov->iov_len || + (char *)sym->PathBuffer + print_offs + print_len > + (char *)iov->iov_base + iov->iov_len) + return -EINVAL; + + return smb2_parse_native_symlink(path, + (char *)sym->PathBuffer + sub_offs, + sub_len, + le32_to_cpu(sym->Flags) & SYMLINK_FLAG_RELATIVE, + full_path, + cifs_sb); +} + +int smb2_open_file(const unsigned int xid, struct cifs_open_parms *oparms, + __u32 *oplock, void *buf) +{ + int rc; + __le16 *smb2_path; + __u8 smb2_oplock; + struct cifs_open_info_data *data = buf; + struct kvec err_iov = {}; + int err_buftype = CIFS_NO_BUFFER; + struct cifs_fid *fid = oparms->fid; + struct network_resiliency_req nr_ioctl_req; + bool retry_without_read_attributes = false; + + smb2_path = cifs_convert_path_to_utf16(oparms->path, oparms->cifs_sb); + if (smb2_path == NULL) + return -ENOMEM; + + /* + * GENERIC_READ, GENERIC_EXECUTE, GENERIC_ALL and MAXIMUM_ALLOWED + * contains also FILE_READ_ATTRIBUTES access right. So do not append + * FILE_READ_ATTRIBUTES when not needed and prevent calling code path + * for retry_without_read_attributes. + */ + if (!(oparms->desired_access & FILE_READ_ATTRIBUTES) && + !(oparms->desired_access & GENERIC_READ) && + !(oparms->desired_access & GENERIC_EXECUTE) && + !(oparms->desired_access & GENERIC_ALL) && + !(oparms->desired_access & MAXIMUM_ALLOWED)) { + oparms->desired_access |= FILE_READ_ATTRIBUTES; + retry_without_read_attributes = true; + } + smb2_oplock = SMB2_OPLOCK_LEVEL_BATCH; + + rc = SMB2_open(xid, oparms, smb2_path, &smb2_oplock, data, NULL, &err_iov, + &err_buftype); + if (rc == -EACCES && retry_without_read_attributes) { + free_rsp_buf(err_buftype, err_iov.iov_base); + memset(&err_iov, 0, sizeof(err_iov)); + err_buftype = CIFS_NO_BUFFER; + oparms->desired_access &= ~FILE_READ_ATTRIBUTES; + rc = SMB2_open(xid, oparms, smb2_path, &smb2_oplock, data, NULL, &err_iov, + &err_buftype); + } + if (rc && data) { + struct smb2_hdr *hdr = err_iov.iov_base; + + if (unlikely(!err_iov.iov_base || err_buftype == CIFS_NO_BUFFER)) + goto out; + if (hdr->Status == STATUS_STOPPED_ON_SYMLINK) { + rc = smb2_parse_symlink_response(oparms->cifs_sb, &err_iov, + oparms->path, + &data->symlink_target); + /* + * If smb2_parse_symlink_response returned -ENODATA then the + * symlink_target was not sent. Treat this as if the SMB2_open() + * failed with STATUS_IO_REPARSE_TAG_NOT_HANDLED status, which is + * indicated by the -EIO errno. + */ + if (rc == -ENODATA) + rc = -EIO; + if (!rc) { + memset(&data->fi, 0, sizeof(data->fi)); + oparms->create_options |= OPEN_REPARSE_POINT; + rc = SMB2_open(xid, oparms, smb2_path, &smb2_oplock, data, + NULL, NULL, NULL); + oparms->create_options &= ~OPEN_REPARSE_POINT; + } + if (!rc) { + bool directory = le32_to_cpu(data->fi.Attributes) & ATTR_DIRECTORY; + rc = smb2_fix_symlink_target_type(&data->symlink_target, + directory, oparms->cifs_sb); + } + } + } + + if (rc) + goto out; + + if (oparms->tcon->use_resilient) { + /* default timeout is 0, servers pick default (120 seconds) */ + nr_ioctl_req.Timeout = + cpu_to_le32(oparms->tcon->handle_timeout); + nr_ioctl_req.Reserved = 0; + rc = SMB2_ioctl(xid, oparms->tcon, fid->persistent_fid, + fid->volatile_fid, FSCTL_LMR_REQUEST_RESILIENCY, + (char *)&nr_ioctl_req, sizeof(nr_ioctl_req), + CIFSMaxBufSize, NULL, NULL /* no return info */); + if (rc == -EOPNOTSUPP) { + cifs_dbg(VFS, + "resiliency not supported by server, disabling\n"); + oparms->tcon->use_resilient = false; + } else if (rc) + cifs_dbg(FYI, "error %d setting resiliency\n", rc); + + rc = 0; + } + + if (data) { + /* if open response does not have IndexNumber field - get it */ + if (data->fi.IndexNumber == 0) { + rc = SMB2_get_srv_num(xid, oparms->tcon, + fid->persistent_fid, + fid->volatile_fid, + &data->fi.IndexNumber); + if (rc) { + /* + * let get_inode_info disable server inode + * numbers + */ + data->fi.IndexNumber = 0; + rc = 0; + } + } + } + + *oplock = smb2_oplock; +out: + free_rsp_buf(err_buftype, err_iov.iov_base); + kfree(smb2_path); + return rc; +} + +int +smb2_unlock_range(struct cifsFileInfo *cfile, struct file_lock *flock, + const unsigned int xid) +{ + int rc = 0, stored_rc; + unsigned int max_num, num = 0, max_buf; + struct smb2_lock_element *buf, *cur; + struct cifs_tcon *tcon = tlink_tcon(cfile->tlink); + struct cifsInodeInfo *cinode = CIFS_I(d_inode(cfile->dentry)); + struct cifsLockInfo *li, *tmp; + __u64 length = 1 + flock->fl_end - flock->fl_start; + LIST_HEAD(tmp_llist); + + /* + * Accessing maxBuf is racy with cifs_reconnect - need to store value + * and check it before using. + */ + max_buf = tcon->ses->server->maxBuf; + if (max_buf < sizeof(struct smb2_lock_element)) + return -EINVAL; + + BUILD_BUG_ON(sizeof(struct smb2_lock_element) > PAGE_SIZE); + max_buf = min_t(unsigned int, max_buf, PAGE_SIZE); + max_num = max_buf / sizeof(struct smb2_lock_element); + buf = kzalloc_objs(struct smb2_lock_element, max_num); + if (!buf) + return -ENOMEM; + + cur = buf; + + cifs_down_write(&cinode->lock_sem); + list_for_each_entry_safe(li, tmp, &cfile->llist->locks, llist) { + if (flock->fl_start > li->offset || + (flock->fl_start + length) < + (li->offset + li->length)) + continue; + if (current->tgid != li->pid) + /* + * flock and OFD lock are associated with an open + * file description, not the process. + */ + if (!(flock->c.flc_flags & (FL_FLOCK | FL_OFDLCK))) + continue; + if (cinode->can_cache_brlcks) { + /* + * We can cache brlock requests - simply remove a lock + * from the file's list. + */ + list_del(&li->llist); + cifs_del_lock_waiters(li); + kfree(li); + continue; + } + cur->Length = cpu_to_le64(li->length); + cur->Offset = cpu_to_le64(li->offset); + cur->Flags = cpu_to_le32(SMB2_LOCKFLAG_UNLOCK); + /* + * We need to save a lock here to let us add it again to the + * file's list if the unlock range request fails on the server. + */ + list_move(&li->llist, &tmp_llist); + if (++num == max_num) { + stored_rc = smb2_lockv(xid, tcon, + cfile->fid.persistent_fid, + cfile->fid.volatile_fid, + current->tgid, num, buf); + if (stored_rc) { + /* + * We failed on the unlock range request - add + * all locks from the tmp list to the head of + * the file's list. + */ + cifs_move_llist(&tmp_llist, + &cfile->llist->locks); + rc = stored_rc; + } else + /* + * The unlock range request succeed - free the + * tmp list. + */ + cifs_free_llist(&tmp_llist); + cur = buf; + num = 0; + } else + cur++; + } + if (num) { + stored_rc = smb2_lockv(xid, tcon, cfile->fid.persistent_fid, + cfile->fid.volatile_fid, current->tgid, + num, buf); + if (stored_rc) { + cifs_move_llist(&tmp_llist, &cfile->llist->locks); + rc = stored_rc; + } else + cifs_free_llist(&tmp_llist); + } + up_write(&cinode->lock_sem); + + kfree(buf); + return rc; +} + +static int +smb2_push_mand_fdlocks(struct cifs_fid_locks *fdlocks, const unsigned int xid, + struct smb2_lock_element *buf, unsigned int max_num) +{ + int rc = 0, stored_rc; + struct cifsFileInfo *cfile = fdlocks->cfile; + struct cifsLockInfo *li; + unsigned int num = 0; + struct smb2_lock_element *cur = buf; + struct cifs_tcon *tcon = tlink_tcon(cfile->tlink); + + list_for_each_entry(li, &fdlocks->locks, llist) { + cur->Length = cpu_to_le64(li->length); + cur->Offset = cpu_to_le64(li->offset); + cur->Flags = cpu_to_le32(li->type | + SMB2_LOCKFLAG_FAIL_IMMEDIATELY); + if (++num == max_num) { + stored_rc = smb2_lockv(xid, tcon, + cfile->fid.persistent_fid, + cfile->fid.volatile_fid, + current->tgid, num, buf); + if (stored_rc) + rc = stored_rc; + cur = buf; + num = 0; + } else + cur++; + } + if (num) { + stored_rc = smb2_lockv(xid, tcon, + cfile->fid.persistent_fid, + cfile->fid.volatile_fid, + current->tgid, num, buf); + if (stored_rc) + rc = stored_rc; + } + + return rc; +} + +int +smb2_push_mandatory_locks(struct cifsFileInfo *cfile) +{ + int rc = 0, stored_rc; + unsigned int xid; + unsigned int max_num, max_buf; + struct smb2_lock_element *buf; + struct cifsInodeInfo *cinode = CIFS_I(d_inode(cfile->dentry)); + struct cifs_fid_locks *fdlocks; + + xid = get_xid(); + + /* + * Accessing maxBuf is racy with cifs_reconnect - need to store value + * and check it for zero before using. + */ + max_buf = tlink_tcon(cfile->tlink)->ses->server->maxBuf; + if (max_buf < sizeof(struct smb2_lock_element)) { + free_xid(xid); + return -EINVAL; + } + + BUILD_BUG_ON(sizeof(struct smb2_lock_element) > PAGE_SIZE); + max_buf = min_t(unsigned int, max_buf, PAGE_SIZE); + max_num = max_buf / sizeof(struct smb2_lock_element); + buf = kzalloc_objs(struct smb2_lock_element, max_num); + if (!buf) { + free_xid(xid); + return -ENOMEM; + } + + list_for_each_entry(fdlocks, &cinode->llist, llist) { + stored_rc = smb2_push_mand_fdlocks(fdlocks, xid, buf, max_num); + if (stored_rc) + rc = stored_rc; + } + + kfree(buf); + free_xid(xid); + return rc; +} -- cgit v1.3.1