projects / packages / kernel.git / blame
| Commit | Line | Data |
|---|---|---|
| 08aa9d92 | 1 | --- linux-2.6.33/scripts/mod/modpost.c~ 2010-02-24 19:52:17.000000000 +0100 |
| 2 | +++ linux-2.6.33/scripts/mod/modpost.c 2010-03-07 14:26:47.242168558 +0100 | |
| 3 | @@ -15,7 +15,8 @@ | |
| 4 | #include <stdio.h> | |
| 5 | #include <ctype.h> | |
| 6 | #include "modpost.h" | |
| 7 | -#include "../../include/generated/autoconf.h" | |
| 8 | +// PLD architectures don't use CONFIG_SYMBOL_PREFIX | |
| 9 | +//#include "../../include/generated/autoconf.h" | |
| 10 | #include "../../include/linux/license.h" | |
| 11 | ||
| 12 | /* Some toolchains use a `_' prefix for all user symbols. */ | |
| 13 | ||
| 2136e199 AM |
14 | --- linux-3.0/scripts/kconfig/lxdialog/check-lxdialog.sh~ 2011-07-22 04:17:23.000000000 +0200 |
| 15 | +++ linux-3.0/scripts/kconfig/lxdialog/check-lxdialog.sh 2011-08-25 21:26:04.799150642 +0200 | |
| 16 | @@ -9,6 +9,12 @@ | |
| 17 | $cc -print-file-name=lib${lib}.${ext} | grep -q / | |
| 18 | if [ $? -eq 0 ]; then | |
| 19 | echo "-l${lib}" | |
| 20 | + for libt in tinfow tinfo ; do | |
| 21 | + $cc -print-file-name=lib${libt}.${ext} | grep -q / | |
| 22 | + if [ $? -eq 0 ]; then | |
| 23 | + echo "-l${libt}" | |
| 24 | + fi | |
| 25 | + done | |
| 26 | exit | |
| 27 | fi | |
| 28 | done | |
| 746b5ec8 | 29 | |
| 08da2677 AM |
30 | From: Eric Sandeen <sandeen@redhat.com> |
| 31 | Subject: [PATCH] xfs: don't wrap ID in xfs_dq_get_next_id | |
| 32 | Message-ID: <f52de68d-abe6-1960-c0ef-1d199346f689@redhat.com> | |
| 33 | Date: Fri, 16 Dec 2016 18:05:20 -0600 | |
| 34 | ||
| 35 | The GETNEXTQOTA ioctl takes whatever ID is sent in, | |
| 36 | and looks for the next active quota for an user | |
| 37 | equal or higher to that ID. | |
| 38 | ||
| 39 | But if we are at the maximum ID and then ask for the "next" | |
| 40 | one, we may wrap back to zero. In this case, userspace | |
| 41 | may loop forever, because it will start querying again | |
| 42 | at zero. | |
| 43 | ||
| 44 | We'll fix this in userspace as well, but for the kernel, | |
| 45 | return -ENOENT if we ask for the next quota ID | |
| 46 | past UINT_MAX so the caller knows to stop. | |
| 47 | ||
| 48 | Signed-off-by: Eric Sandeen <sandeen@redhat.com> | |
| 49 | --- | |
| 50 | ||
| 51 | diff --git a/fs/xfs/xfs_dquot.c b/fs/xfs/xfs_dquot.c | |
| 52 | index 7a30b8f..dbeddd9 100644 | |
| 53 | --- a/fs/xfs/xfs_dquot.c | |
| 54 | +++ b/fs/xfs/xfs_dquot.c | |
| 55 | @@ -710,6 +710,10 @@ | |
| 56 | /* Simple advance */ | |
| 57 | next_id = *id + 1; | |
| 58 | ||
| 59 | + /* If we'd wrap past the max ID, stop */ | |
| 60 | + if (next_id < *id) | |
| 61 | + return -ENOENT; | |
| 62 | + | |
| 63 | /* If new ID is within the current chunk, advancing it sufficed */ | |
| 64 | if (next_id % mp->m_quotainfo->qi_dqperchunk) { | |
| 65 | *id = next_id; | |
| 66 | ||
| 53badc7f AM |
67 | From: Brian Foster <bfoster@redhat.com> |
| 68 | Subject: [PATCH] xfs: prevent quotacheck from overloading inode lru | |
| 69 | ||
| 70 | Quotacheck runs at mount time in situations where quota accounting must | |
| 71 | be recalculated. In doing so, it uses bulkstat to visit every inode in | |
| 72 | the filesystem. Historically, every inode processed during quotacheck | |
| 73 | was released and immediately tagged for reclaim because quotacheck runs | |
| 74 | before the superblock is marked active by the VFS. In other words, | |
| 75 | the final iput() lead to an immediate ->destroy_inode() call, which | |
| 76 | allowed the XFS background reclaim worker to start reclaiming inodes. | |
| 77 | ||
| 78 | Commit 17c12bcd3 ("xfs: when replaying bmap operations, don't let | |
| 79 | unlinked inodes get reaped") marks the XFS superblock active sooner as | |
| 80 | part of the mount process to support caching inodes processed during log | |
| 81 | recovery. This occurs before quotacheck and thus means all inodes | |
| 82 | processed by quotacheck are inserted to the LRU on release. The | |
| 83 | s_umount lock is held until the mount has completed and thus prevents | |
| 84 | the shrinkers from operating on the sb. This means that quotacheck can | |
| 85 | excessively populate the inode LRU and lead to OOM conditions on systems | |
| 86 | without sufficient RAM. | |
| 87 | ||
| 88 | Update the quotacheck bulkstat handler to set XFS_IGET_DONTCACHE on | |
| 89 | inodes processed by quotacheck. This causes ->drop_inode() to return 1 | |
| 90 | and in turn causes iput_final() to evict the inode. This preserves the | |
| 91 | original quotacheck behavior and prevents it from overloading the LRU | |
| 92 | and running out of memory. | |
| 93 | ||
| 94 | CC: stable@vger.kernel.org # v4.9 | |
| 95 | Reported-by: Martin Svec <martin.svec@zoner.cz> | |
| 96 | Signed-off-by: Brian Foster <bfoster@redhat.com> | |
| 97 | --- | |
| 98 | fs/xfs/xfs_qm.c | 3 ++- | |
| 99 | 1 file changed, 2 insertions(+), 1 deletion(-) | |
| 100 | ||
| 101 | diff --git a/fs/xfs/xfs_qm.c b/fs/xfs/xfs_qm.c | |
| 102 | index 45e50ea..b669b12 100644 | |
| 103 | --- a/fs/xfs/xfs_qm.c | |
| 104 | +++ b/fs/xfs/xfs_qm.c | |
| 105 | @@ -1177,7 +1177,8 @@ xfs_qm_dqusage_adjust( | |
| 106 | * the case in all other instances. It's OK that we do this because | |
| 107 | * quotacheck is done only at mount time. | |
| 108 | */ | |
| 109 | - error = xfs_iget(mp, NULL, ino, 0, XFS_ILOCK_EXCL, &ip); | |
| 110 | + error = xfs_iget(mp, NULL, ino, XFS_IGET_DONTCACHE, XFS_ILOCK_EXCL, | |
| 111 | + &ip); | |
| 112 | if (error) { | |
| 113 | *res = BULKSTAT_RV_NOTHING; | |
| 114 | return error; | |
| 115 | -- | |
| 116 | 2.7.4 | |
| 117 |