[people/ms/linux.git] / fs / xfs / scrub / health.c

// SPDX-License-Identifier: GPL-2.0+
/*
 * Copyright (C) 2019 Oracle.  All Rights Reserved.
 * Author: Darrick J. Wong <darrick.wong@oracle.com>
 */
#include "xfs.h"
#include "xfs_fs.h"
#include "xfs_shared.h"
#include "xfs_format.h"
#include "xfs_btree.h"
#include "xfs_trans_resv.h"
#include "xfs_mount.h"
#include "xfs_ag.h"
#include "xfs_health.h"
#include "scrub/scrub.h"
#include "scrub/health.h"

/*
 * Scrub and In-Core Filesystem Health Assessments
 * ===============================================
 *
 * Online scrub and repair have the time and the ability to perform stronger
 * checks than we can do from the metadata verifiers, because they can
 * cross-reference records between data structures.  Therefore, scrub is in a
 * good position to update the online filesystem health assessments to reflect
 * the good/bad state of the data structure.
 *
 * We therefore extend scrub in the following ways to achieve this:
 *
 * 1. Create a "sick_mask" field in the scrub context.  When we're setting up a
 * scrub call, set this to the default XFS_SICK_* flag(s) for the selected
 * scrub type (call it A).  Scrub and repair functions can override the default
 * sick_mask value if they choose.
 *
 * 2. If the scrubber returns a runtime error code, we exit making no changes
 * to the incore sick state.
 *
 * 3. If the scrubber finds that A is clean, use sick_mask to clear the incore
 * sick flags before exiting.
 *
 * 4. If the scrubber finds that A is corrupt, use sick_mask to set the incore
 * sick flags.  If the user didn't want to repair then we exit, leaving the
 * metadata structure unfixed and the sick flag set.
 *
 * 5. Now we know that A is corrupt and the user wants to repair, so run the
 * repairer.  If the repairer returns an error code, we exit with that error
 * code, having made no further changes to the incore sick state.
 *
 * 6. If repair rebuilds A correctly and the subsequent re-scrub of A is clean,
 * use sick_mask to clear the incore sick flags.  This should have the effect
 * that A is no longer marked sick.
 *
 * 7. If repair rebuilds A incorrectly, the re-scrub will find it corrupt and
 * use sick_mask to set the incore sick flags.  This should have no externally
 * visible effect since we already set them in step (4).
 *
 * There are some complications to this story, however.  For certain types of
 * complementary metadata indices (e.g. inobt/finobt), it is easier to rebuild
 * both structures at the same time.  The following principles apply to this
 * type of repair strategy:
 *
 * 8. Any repair function that rebuilds multiple structures should update
 * sick_mask_visible to reflect whatever other structures are rebuilt, and
 * verify that all the rebuilt structures can pass a scrub check.  The outcomes
 * of 5-7 still apply, but with a sick_mask that covers everything being
 * rebuilt.
 */

/* Map our scrub type to a sick mask and a set of health update functions. */

enum xchk_health_group {
	XHG_FS = 1,
	XHG_RT,
	XHG_AG,
	XHG_INO,
};

struct xchk_health_map {
	enum xchk_health_group	group;
	unsigned int		sick_mask;
};

static const struct xchk_health_map type_to_health_flag[XFS_SCRUB_TYPE_NR] = {
	[XFS_SCRUB_TYPE_SB]		= { XHG_AG,  XFS_SICK_AG_SB },
	[XFS_SCRUB_TYPE_AGF]		= { XHG_AG,  XFS_SICK_AG_AGF },
	[XFS_SCRUB_TYPE_AGFL]		= { XHG_AG,  XFS_SICK_AG_AGFL },
	[XFS_SCRUB_TYPE_AGI]		= { XHG_AG,  XFS_SICK_AG_AGI },
	[XFS_SCRUB_TYPE_BNOBT]		= { XHG_AG,  XFS_SICK_AG_BNOBT },
	[XFS_SCRUB_TYPE_CNTBT]		= { XHG_AG,  XFS_SICK_AG_CNTBT },
	[XFS_SCRUB_TYPE_INOBT]		= { XHG_AG,  XFS_SICK_AG_INOBT },
	[XFS_SCRUB_TYPE_FINOBT]		= { XHG_AG,  XFS_SICK_AG_FINOBT },
	[XFS_SCRUB_TYPE_RMAPBT]		= { XHG_AG,  XFS_SICK_AG_RMAPBT },
	[XFS_SCRUB_TYPE_REFCNTBT]	= { XHG_AG,  XFS_SICK_AG_REFCNTBT },
	[XFS_SCRUB_TYPE_INODE]		= { XHG_INO, XFS_SICK_INO_CORE },
	[XFS_SCRUB_TYPE_BMBTD]		= { XHG_INO, XFS_SICK_INO_BMBTD },
	[XFS_SCRUB_TYPE_BMBTA]		= { XHG_INO, XFS_SICK_INO_BMBTA },
	[XFS_SCRUB_TYPE_BMBTC]		= { XHG_INO, XFS_SICK_INO_BMBTC },
	[XFS_SCRUB_TYPE_DIR]		= { XHG_INO, XFS_SICK_INO_DIR },
	[XFS_SCRUB_TYPE_XATTR]		= { XHG_INO, XFS_SICK_INO_XATTR },
	[XFS_SCRUB_TYPE_SYMLINK]	= { XHG_INO, XFS_SICK_INO_SYMLINK },
	[XFS_SCRUB_TYPE_PARENT]		= { XHG_INO, XFS_SICK_INO_PARENT },
	[XFS_SCRUB_TYPE_RTBITMAP]	= { XHG_RT,  XFS_SICK_RT_BITMAP },
	[XFS_SCRUB_TYPE_RTSUM]		= { XHG_RT,  XFS_SICK_RT_SUMMARY },
	[XFS_SCRUB_TYPE_UQUOTA]		= { XHG_FS,  XFS_SICK_FS_UQUOTA },
	[XFS_SCRUB_TYPE_GQUOTA]		= { XHG_FS,  XFS_SICK_FS_GQUOTA },
	[XFS_SCRUB_TYPE_PQUOTA]		= { XHG_FS,  XFS_SICK_FS_PQUOTA },
	[XFS_SCRUB_TYPE_FSCOUNTERS]	= { XHG_FS,  XFS_SICK_FS_COUNTERS },
};

/* Return the health status mask for this scrub type. */
unsigned int
xchk_health_mask_for_scrub_type(
	__u32			scrub_type)
{
	return type_to_health_flag[scrub_type].sick_mask;
}

/*
 * Update filesystem health assessments based on what we found and did.
 *
 * If the scrubber finds errors, we mark sick whatever's mentioned in
 * sick_mask, no matter whether this is a first scan or an
 * evaluation of repair effectiveness.
 *
 * Otherwise, no direct corruption was found, so mark whatever's in
 * sick_mask as healthy.
 */
void
xchk_update_health(
	struct xfs_scrub	*sc)
{
	struct xfs_perag	*pag;
	bool			bad;

	if (!sc->sick_mask)
		return;

	bad = (sc->sm->sm_flags & (XFS_SCRUB_OFLAG_CORRUPT |
				   XFS_SCRUB_OFLAG_XCORRUPT));
	switch (type_to_health_flag[sc->sm->sm_type].group) {
	case XHG_AG:
		pag = xfs_perag_get(sc->mp, sc->sm->sm_agno);
		if (bad)
			xfs_ag_mark_sick(pag, sc->sick_mask);
		else
			xfs_ag_mark_healthy(pag, sc->sick_mask);
		xfs_perag_put(pag);
		break;
	case XHG_INO:
		if (!sc->ip)
			return;
		if (bad)
			xfs_inode_mark_sick(sc->ip, sc->sick_mask);
		else
			xfs_inode_mark_healthy(sc->ip, sc->sick_mask);
		break;
	case XHG_FS:
		if (bad)
			xfs_fs_mark_sick(sc->mp, sc->sick_mask);
		else
			xfs_fs_mark_healthy(sc->mp, sc->sick_mask);
		break;
	case XHG_RT:
		if (bad)
			xfs_rt_mark_sick(sc->mp, sc->sick_mask);
		else
			xfs_rt_mark_healthy(sc->mp, sc->sick_mask);
		break;
	default:
		ASSERT(0);
		break;
	}
}

/* Is the given per-AG btree healthy enough for scanning? */
bool
xchk_ag_btree_healthy_enough(
	struct xfs_scrub	*sc,
	struct xfs_perag	*pag,
	xfs_btnum_t		btnum)
{
	unsigned int		mask = 0;

	/*
	 * We always want the cursor if it's the same type as whatever we're
	 * scrubbing, even if we already know the structure is corrupt.
	 *
	 * Otherwise, we're only interested in the btree for cross-referencing.
	 * If we know the btree is bad then don't bother, just set XFAIL.
	 */
	switch (btnum) {
	case XFS_BTNUM_BNO:
		if (sc->sm->sm_type == XFS_SCRUB_TYPE_BNOBT)
			return true;
		mask = XFS_SICK_AG_BNOBT;
		break;
	case XFS_BTNUM_CNT:
		if (sc->sm->sm_type == XFS_SCRUB_TYPE_CNTBT)
			return true;
		mask = XFS_SICK_AG_CNTBT;
		break;
	case XFS_BTNUM_INO:
		if (sc->sm->sm_type == XFS_SCRUB_TYPE_INOBT)
			return true;
		mask = XFS_SICK_AG_INOBT;
		break;
	case XFS_BTNUM_FINO:
		if (sc->sm->sm_type == XFS_SCRUB_TYPE_FINOBT)
			return true;
		mask = XFS_SICK_AG_FINOBT;
		break;
	case XFS_BTNUM_RMAP:
		if (sc->sm->sm_type == XFS_SCRUB_TYPE_RMAPBT)
			return true;
		mask = XFS_SICK_AG_RMAPBT;
		break;
	case XFS_BTNUM_REFC:
		if (sc->sm->sm_type == XFS_SCRUB_TYPE_REFCNTBT)
			return true;
		mask = XFS_SICK_AG_REFCNTBT;
		break;
	default:
		ASSERT(0);
		return true;
	}

	if (xfs_ag_has_sickness(pag, mask)) {
		sc->sm->sm_flags |= XFS_SCRUB_OFLAG_XFAIL;
		return false;
	}

	return true;
}
Commit	Line	Data
4860a05d DW	1	// SPDX-License-Identifier: GPL-2.0+
	2	/*
	3	* Copyright (C) 2019 Oracle. All Rights Reserved.
	4	* Author: Darrick J. Wong <darrick.wong@oracle.com>
	5	*/
	6	#include "xfs.h"
	7	#include "xfs_fs.h"
	8	#include "xfs_shared.h"
	9	#include "xfs_format.h"
4860a05d	10	#include "xfs_btree.h"
36029dee DC	11	#include "xfs_trans_resv.h"
36029dee DC	12	#include "xfs_mount.h"
9bbafc71	13	#include "xfs_ag.h"
4860a05d DW	14	#include "xfs_health.h"
4860a05d DW	15	#include "scrub/scrub.h"
5f213ddb	16	#include "scrub/health.h"
4860a05d DW	17
	18	/*
	19	* Scrub and In-Core Filesystem Health Assessments
	20	* ===============================================
	21	*
	22	* Online scrub and repair have the time and the ability to perform stronger
	23	* checks than we can do from the metadata verifiers, because they can
	24	* cross-reference records between data structures. Therefore, scrub is in a
	25	* good position to update the online filesystem health assessments to reflect
	26	* the good/bad state of the data structure.
	27	*
	28	* We therefore extend scrub in the following ways to achieve this:
	29	*
	30	* 1. Create a "sick_mask" field in the scrub context. When we're setting up a
	31	* scrub call, set this to the default XFS_SICK_* flag(s) for the selected
	32	* scrub type (call it A). Scrub and repair functions can override the default
	33	* sick_mask value if they choose.
	34	*
	35	* 2. If the scrubber returns a runtime error code, we exit making no changes
	36	* to the incore sick state.
	37	*
	38	* 3. If the scrubber finds that A is clean, use sick_mask to clear the incore
	39	* sick flags before exiting.
	40	*
	41	* 4. If the scrubber finds that A is corrupt, use sick_mask to set the incore
	42	* sick flags. If the user didn't want to repair then we exit, leaving the
	43	* metadata structure unfixed and the sick flag set.
	44	*
	45	* 5. Now we know that A is corrupt and the user wants to repair, so run the
	46	* repairer. If the repairer returns an error code, we exit with that error
	47	* code, having made no further changes to the incore sick state.
	48	*
	49	* 6. If repair rebuilds A correctly and the subsequent re-scrub of A is clean,
	50	* use sick_mask to clear the incore sick flags. This should have the effect
	51	* that A is no longer marked sick.
	52	*
	53	* 7. If repair rebuilds A incorrectly, the re-scrub will find it corrupt and
	54	* use sick_mask to set the incore sick flags. This should have no externally
	55	* visible effect since we already set them in step (4).
	56	*
	57	* There are some complications to this story, however. For certain types of
	58	* complementary metadata indices (e.g. inobt/finobt), it is easier to rebuild
	59	* both structures at the same time. The following principles apply to this
	60	* type of repair strategy:
	61	*
	62	* 8. Any repair function that rebuilds multiple structures should update
	63	* sick_mask_visible to reflect whatever other structures are rebuilt, and
	64	* verify that all the rebuilt structures can pass a scrub check. The outcomes
	65	* of 5-7 still apply, but with a sick_mask that covers everything being
	66	* rebuilt.
	67	*/
	68
	69	/* Map our scrub type to a sick mask and a set of health update functions. */
	70
	71	enum xchk_health_group {
	72	XHG_FS = 1,
	73	XHG_RT,
	74	XHG_AG,
	75	XHG_INO,
	76	};
	77
	78	struct xchk_health_map {
	79	enum xchk_health_group group;
	80	unsigned int sick_mask;
81	};
82
83	static const struct xchk_health_map type_to_health_flag[XFS_SCRUB_TYPE_NR] = {
84	[XFS_SCRUB_TYPE_SB] = { XHG_AG, XFS_SICK_AG_SB },
85	[XFS_SCRUB_TYPE_AGF] = { XHG_AG, XFS_SICK_AG_AGF },
86	[XFS_SCRUB_TYPE_AGFL] = { XHG_AG, XFS_SICK_AG_AGFL },
87	[XFS_SCRUB_TYPE_AGI] = { XHG_AG, XFS_SICK_AG_AGI },
88	[XFS_SCRUB_TYPE_BNOBT] = { XHG_AG, XFS_SICK_AG_BNOBT },
89	[XFS_SCRUB_TYPE_CNTBT] = { XHG_AG, XFS_SICK_AG_CNTBT },
90	[XFS_SCRUB_TYPE_INOBT] = { XHG_AG, XFS_SICK_AG_INOBT },
91	[XFS_SCRUB_TYPE_FINOBT] = { XHG_AG, XFS_SICK_AG_FINOBT },
92	[XFS_SCRUB_TYPE_RMAPBT] = { XHG_AG, XFS_SICK_AG_RMAPBT },
93	[XFS_SCRUB_TYPE_REFCNTBT] = { XHG_AG, XFS_SICK_AG_REFCNTBT },
94	[XFS_SCRUB_TYPE_INODE] = { XHG_INO, XFS_SICK_INO_CORE },
95	[XFS_SCRUB_TYPE_BMBTD] = { XHG_INO, XFS_SICK_INO_BMBTD },
96	[XFS_SCRUB_TYPE_BMBTA] = { XHG_INO, XFS_SICK_INO_BMBTA },
97	[XFS_SCRUB_TYPE_BMBTC] = { XHG_INO, XFS_SICK_INO_BMBTC },
98	[XFS_SCRUB_TYPE_DIR] = { XHG_INO, XFS_SICK_INO_DIR },
99	[XFS_SCRUB_TYPE_XATTR] = { XHG_INO, XFS_SICK_INO_XATTR },
100	[XFS_SCRUB_TYPE_SYMLINK] = { XHG_INO, XFS_SICK_INO_SYMLINK },
101	[XFS_SCRUB_TYPE_PARENT] = { XHG_INO, XFS_SICK_INO_PARENT },
102	[XFS_SCRUB_TYPE_RTBITMAP] = { XHG_RT, XFS_SICK_RT_BITMAP },
103	[XFS_SCRUB_TYPE_RTSUM] = { XHG_RT, XFS_SICK_RT_SUMMARY },
104	[XFS_SCRUB_TYPE_UQUOTA] = { XHG_FS, XFS_SICK_FS_UQUOTA },
105	[XFS_SCRUB_TYPE_GQUOTA] = { XHG_FS, XFS_SICK_FS_GQUOTA },
106	[XFS_SCRUB_TYPE_PQUOTA] = { XHG_FS, XFS_SICK_FS_PQUOTA },
75efa57d	107	[XFS_SCRUB_TYPE_FSCOUNTERS] = { XHG_FS, XFS_SICK_FS_COUNTERS },
4860a05d DW	108	};
	109
	110	/* Return the health status mask for this scrub type. */
	111	unsigned int
	112	xchk_health_mask_for_scrub_type(
	113	__u32 scrub_type)
	114	{
	115	return type_to_health_flag[scrub_type].sick_mask;
	116	}
	117
	118	/*
	119	* Update filesystem health assessments based on what we found and did.
	120	*
	121	* If the scrubber finds errors, we mark sick whatever's mentioned in
	122	* sick_mask, no matter whether this is a first scan or an
	123	* evaluation of repair effectiveness.
	124	*
	125	* Otherwise, no direct corruption was found, so mark whatever's in
	126	* sick_mask as healthy.
	127	*/
	128	void
	129	xchk_update_health(
	130	struct xfs_scrub *sc)
	131	{
	132	struct xfs_perag *pag;
	133	bool bad;
	134
	135	if (!sc->sick_mask)
	136	return;
	137
9de4b514 DW	138	bad = (sc->sm->sm_flags & (XFS_SCRUB_OFLAG_CORRUPT \|
9de4b514 DW	139	XFS_SCRUB_OFLAG_XCORRUPT));
4860a05d DW	140	switch (type_to_health_flag[sc->sm->sm_type].group) {
	141	case XHG_AG:
	142	pag = xfs_perag_get(sc->mp, sc->sm->sm_agno);
	143	if (bad)
	144	xfs_ag_mark_sick(pag, sc->sick_mask);
	145	else
	146	xfs_ag_mark_healthy(pag, sc->sick_mask);
	147	xfs_perag_put(pag);
	148	break;
	149	case XHG_INO:
	150	if (!sc->ip)
	151	return;
	152	if (bad)
	153	xfs_inode_mark_sick(sc->ip, sc->sick_mask);
	154	else
	155	xfs_inode_mark_healthy(sc->ip, sc->sick_mask);
	156	break;
	157	case XHG_FS:
	158	if (bad)
	159	xfs_fs_mark_sick(sc->mp, sc->sick_mask);
	160	else
	161	xfs_fs_mark_healthy(sc->mp, sc->sick_mask);
	162	break;
	163	case XHG_RT:
	164	if (bad)
	165	xfs_rt_mark_sick(sc->mp, sc->sick_mask);
	166	else
	167	xfs_rt_mark_healthy(sc->mp, sc->sick_mask);
	168	break;
	169	default:
	170	ASSERT(0);
	171	break;
	172	}
	173	}
4fb7951f DW	174
	175	/* Is the given per-AG btree healthy enough for scanning? */
	176	bool
	177	xchk_ag_btree_healthy_enough(
	178	struct xfs_scrub *sc,
	179	struct xfs_perag *pag,
	180	xfs_btnum_t btnum)
	181	{
	182	unsigned int mask = 0;
	183
	184	/*
	185	* We always want the cursor if it's the same type as whatever we're
	186	* scrubbing, even if we already know the structure is corrupt.
	187	*
	188	* Otherwise, we're only interested in the btree for cross-referencing.
	189	* If we know the btree is bad then don't bother, just set XFAIL.
	190	*/
	191	switch (btnum) {
	192	case XFS_BTNUM_BNO:
	193	if (sc->sm->sm_type == XFS_SCRUB_TYPE_BNOBT)
	194	return true;
	195	mask = XFS_SICK_AG_BNOBT;
	196	break;
	197	case XFS_BTNUM_CNT:
	198	if (sc->sm->sm_type == XFS_SCRUB_TYPE_CNTBT)
	199	return true;
	200	mask = XFS_SICK_AG_CNTBT;
	201	break;
	202	case XFS_BTNUM_INO:
	203	if (sc->sm->sm_type == XFS_SCRUB_TYPE_INOBT)
	204	return true;
	205	mask = XFS_SICK_AG_INOBT;
	206	break;
	207	case XFS_BTNUM_FINO:
	208	if (sc->sm->sm_type == XFS_SCRUB_TYPE_FINOBT)
	209	return true;
	210	mask = XFS_SICK_AG_FINOBT;
	211	break;
	212	case XFS_BTNUM_RMAP:
	213	if (sc->sm->sm_type == XFS_SCRUB_TYPE_RMAPBT)
	214	return true;
	215	mask = XFS_SICK_AG_RMAPBT;
	216	break;
	217	case XFS_BTNUM_REFC:
	218	if (sc->sm->sm_type == XFS_SCRUB_TYPE_REFCNTBT)
	219	return true;
	220	mask = XFS_SICK_AG_REFCNTBT;
	221	break;
	222	default:
	223	ASSERT(0);
	224	return true;
	225	}
	226
	227	if (xfs_ag_has_sickness(pag, mask)) {
	228	sc->sm->sm_flags \|= XFS_SCRUB_OFLAG_XFAIL;
	229	return false;
	230	}
	231
	232	return true;
	233	}