X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libbcachefs%2Ferror.c;h=d32c8bebe46c32f7abc1a11ad49ee80752f2a623;hb=b5fd066153c40a70a29caa1ea7987723ab687763;hp=ca2a06e2e4eb3dbd751aa24ecb4c1622cc2bb594;hpb=1cf4d51dc4661f336f5318c176a3561ddf5bf04f;p=bcachefs-tools-debian

diff --git a/libbcachefs/error.c b/libbcachefs/error.c
index ca2a06e..d32c8be 100644
--- a/libbcachefs/error.c
+++ b/libbcachefs/error.c
@@ -1,29 +1,41 @@
+// SPDX-License-Identifier: GPL-2.0
 #include "bcachefs.h"
 #include "error.h"
-#include "io.h"
 #include "super.h"
+#include "thread_with_file.h"
 
-void bch2_inconsistent_error(struct bch_fs *c)
+#define FSCK_ERR_RATELIMIT_NR	10
+
+bool bch2_inconsistent_error(struct bch_fs *c)
 {
-	set_bit(BCH_FS_ERROR, &c->flags);
+	set_bit(BCH_FS_error, &c->flags);
 
 	switch (c->opts.errors) {
-	case BCH_ON_ERROR_CONTINUE:
-		break;
-	case BCH_ON_ERROR_RO:
+	case BCH_ON_ERROR_continue:
+		return false;
+	case BCH_ON_ERROR_ro:
 		if (bch2_fs_emergency_read_only(c))
-			bch_err(c, "emergency read only");
-		break;
-	case BCH_ON_ERROR_PANIC:
+			bch_err(c, "inconsistency detected - emergency read only");
+		return true;
+	case BCH_ON_ERROR_panic:
 		panic(bch2_fmt(c, "panic after error"));
-		break;
+		return true;
+	default:
+		BUG();
 	}
 }
 
+void bch2_topology_error(struct bch_fs *c)
+{
+	set_bit(BCH_FS_topology_error, &c->flags);
+	if (!test_bit(BCH_FS_fsck_running, &c->flags))
+		bch2_inconsistent_error(c);
+}
+
 void bch2_fatal_error(struct bch_fs *c)
 {
 	if (bch2_fs_emergency_read_only(c))
-		bch_err(c, "emergency read only");
+		bch_err(c, "fatal error - emergency read only");
 }
 
 void bch2_io_error_work(struct work_struct *work)
@@ -32,125 +44,294 @@ void bch2_io_error_work(struct work_struct *work)
 	struct bch_fs *c = ca->fs;
 	bool dev;
 
-	mutex_lock(&c->state_lock);
-	dev = bch2_dev_state_allowed(c, ca, BCH_MEMBER_STATE_RO,
+	down_write(&c->state_lock);
+	dev = bch2_dev_state_allowed(c, ca, BCH_MEMBER_STATE_ro,
 				    BCH_FORCE_IF_DEGRADED);
 	if (dev
-	    ? __bch2_dev_set_state(c, ca, BCH_MEMBER_STATE_RO,
+	    ? __bch2_dev_set_state(c, ca, BCH_MEMBER_STATE_ro,
 				  BCH_FORCE_IF_DEGRADED)
 	    : bch2_fs_emergency_read_only(c))
 		bch_err(ca,
 			"too many IO errors, setting %s RO",
 			dev ? "device" : "filesystem");
-	mutex_unlock(&c->state_lock);
+	up_write(&c->state_lock);
 }
 
-void bch2_io_error(struct bch_dev *ca)
+void bch2_io_error(struct bch_dev *ca, enum bch_member_error_type type)
 {
+	atomic64_inc(&ca->errors[type]);
 	//queue_work(system_long_wq, &ca->io_error_work);
 }
 
+enum ask_yn {
+	YN_NO,
+	YN_YES,
+	YN_ALLNO,
+	YN_ALLYES,
+};
+
+static enum ask_yn parse_yn_response(char *buf)
+{
+	buf = strim(buf);
+
+	if (strlen(buf) == 1)
+		switch (buf[0]) {
+		case 'n':
+			return YN_NO;
+		case 'y':
+			return YN_YES;
+		case 'N':
+			return YN_ALLNO;
+		case 'Y':
+			return YN_ALLYES;
+		}
+	return -1;
+}
+
 #ifdef __KERNEL__
-#define ask_yn()	false
+static enum ask_yn bch2_fsck_ask_yn(struct bch_fs *c)
+{
+	struct stdio_redirect *stdio = c->stdio;
+
+	if (c->stdio_filter && c->stdio_filter != current)
+		stdio = NULL;
+
+	if (!stdio)
+		return YN_NO;
+
+	char buf[100];
+	int ret;
+
+	do {
+		bch2_print(c, " (y,n, or Y,N for all errors of this type) ");
+
+		int r = bch2_stdio_redirect_readline(stdio, buf, sizeof(buf) - 1);
+		if (r < 0)
+			return YN_NO;
+		buf[r] = '\0';
+	} while ((ret = parse_yn_response(buf)) < 0);
+
+	return ret;
+}
 #else
+
 #include "tools-util.h"
+
+static enum ask_yn bch2_fsck_ask_yn(struct bch_fs *c)
+{
+	char *buf = NULL;
+	size_t buflen = 0;
+	int ret;
+
+	do {
+		fputs(" (y,n, or Y,N for all errors of this type) ", stdout);
+		fflush(stdout);
+
+		if (getline(&buf, &buflen, stdin) < 0)
+			die("error reading from standard input");
+	} while ((ret = parse_yn_response(buf)) < 0);
+
+	free(buf);
+	return ret;
+}
+
 #endif
 
-enum fsck_err_ret bch2_fsck_err(struct bch_fs *c, unsigned flags,
-				const char *fmt, ...)
+static struct fsck_err_state *fsck_err_get(struct bch_fs *c, const char *fmt)
 {
 	struct fsck_err_state *s;
-	va_list args;
-	bool fix = false, print = true, suppressing = false;
-	char _buf[sizeof(s->buf)], *buf = _buf;
-
-	mutex_lock(&c->fsck_error_lock);
 
-	if (test_bit(BCH_FS_FSCK_DONE, &c->flags))
-		goto print;
+	if (!test_bit(BCH_FS_fsck_running, &c->flags))
+		return NULL;
 
-	list_for_each_entry(s, &c->fsck_errors, list)
-		if (s->fmt == fmt)
-			goto found;
+	list_for_each_entry(s, &c->fsck_error_msgs, list)
+		if (s->fmt == fmt) {
+			/*
+			 * move it to the head of the list: repeated fsck errors
+			 * are common
+			 */
+			list_move(&s->list, &c->fsck_error_msgs);
+			return s;
+		}
 
-	s = kzalloc(sizeof(*s), GFP_KERNEL);
+	s = kzalloc(sizeof(*s), GFP_NOFS);
 	if (!s) {
-		if (!c->fsck_alloc_err)
+		if (!c->fsck_alloc_msgs_err)
 			bch_err(c, "kmalloc err, cannot ratelimit fsck errs");
-		c->fsck_alloc_err = true;
-		buf = _buf;
-		goto print;
+		c->fsck_alloc_msgs_err = true;
+		return NULL;
 	}
 
 	INIT_LIST_HEAD(&s->list);
 	s->fmt = fmt;
-found:
-	list_move(&s->list, &c->fsck_errors);
-	s->nr++;
-	suppressing	= s->nr == 10;
-	print		= s->nr <= 10;
-	buf		= s->buf;
-print:
+	list_add(&s->list, &c->fsck_error_msgs);
+	return s;
+}
+
+int bch2_fsck_err(struct bch_fs *c,
+		  enum bch_fsck_flags flags,
+		  enum bch_sb_error_id err,
+		  const char *fmt, ...)
+{
+	struct fsck_err_state *s = NULL;
+	va_list args;
+	bool print = true, suppressing = false, inconsistent = false;
+	struct printbuf buf = PRINTBUF, *out = &buf;
+	int ret = -BCH_ERR_fsck_ignore;
+
+	if ((flags & FSCK_CAN_FIX) &&
+	    test_bit(err, c->sb.errors_silent))
+		return -BCH_ERR_fsck_fix;
+
+	bch2_sb_error_count(c, err);
+
 	va_start(args, fmt);
-	vscnprintf(buf, sizeof(_buf), fmt, args);
+	prt_vprintf(out, fmt, args);
 	va_end(args);
 
-	if (c->opts.fix_errors == FSCK_OPT_EXIT) {
-		bch_err(c, "%s, exiting", buf);
-		mutex_unlock(&c->fsck_error_lock);
-		return FSCK_ERR_EXIT;
+	mutex_lock(&c->fsck_error_msgs_lock);
+	s = fsck_err_get(c, fmt);
+	if (s) {
+		/*
+		 * We may be called multiple times for the same error on
+		 * transaction restart - this memoizes instead of asking the user
+		 * multiple times for the same error:
+		 */
+		if (s->last_msg && !strcmp(buf.buf, s->last_msg)) {
+			ret = s->ret;
+			mutex_unlock(&c->fsck_error_msgs_lock);
+			printbuf_exit(&buf);
+			return ret;
+		}
+
+		kfree(s->last_msg);
+		s->last_msg = kstrdup(buf.buf, GFP_KERNEL);
+
+		if (c->opts.ratelimit_errors &&
+		    !(flags & FSCK_NO_RATELIMIT) &&
+		    s->nr >= FSCK_ERR_RATELIMIT_NR) {
+			if (s->nr == FSCK_ERR_RATELIMIT_NR)
+				suppressing = true;
+			else
+				print = false;
+		}
+
+		s->nr++;
 	}
 
-	if (flags & FSCK_CAN_FIX) {
-		if (c->opts.fix_errors == FSCK_OPT_ASK) {
-			printk(KERN_ERR "%s: fix?", buf);
-			fix = ask_yn();
-		} else if (c->opts.fix_errors == FSCK_OPT_YES ||
+#ifdef BCACHEFS_LOG_PREFIX
+	if (!strncmp(fmt, "bcachefs:", 9))
+		prt_printf(out, bch2_log_msg(c, ""));
+#endif
+
+	if (!test_bit(BCH_FS_fsck_running, &c->flags)) {
+		if (c->opts.errors != BCH_ON_ERROR_continue ||
+		    !(flags & (FSCK_CAN_FIX|FSCK_CAN_IGNORE))) {
+			prt_str(out, ", shutting down");
+			inconsistent = true;
+			ret = -BCH_ERR_fsck_errors_not_fixed;
+		} else if (flags & FSCK_CAN_FIX) {
+			prt_str(out, ", fixing");
+			ret = -BCH_ERR_fsck_fix;
+		} else {
+			prt_str(out, ", continuing");
+			ret = -BCH_ERR_fsck_ignore;
+		}
+	} else if (c->opts.fix_errors == FSCK_FIX_exit) {
+		prt_str(out, ", exiting");
+		ret = -BCH_ERR_fsck_errors_not_fixed;
+	} else if (flags & FSCK_CAN_FIX) {
+		int fix = s && s->fix
+			? s->fix
+			: c->opts.fix_errors;
+
+		if (fix == FSCK_FIX_ask) {
+			int ask;
+
+			prt_str(out, ": fix?");
+			if (bch2_fs_stdio_redirect(c))
+				bch2_print(c, "%s", out->buf);
+			else
+				bch2_print_string_as_lines(KERN_ERR, out->buf);
+			print = false;
+
+			ask = bch2_fsck_ask_yn(c);
+
+			if (ask >= YN_ALLNO && s)
+				s->fix = ask == YN_ALLNO
+					? FSCK_FIX_no
+					: FSCK_FIX_yes;
+
+			ret = ask & 1
+				? -BCH_ERR_fsck_fix
+				: -BCH_ERR_fsck_ignore;
+		} else if (fix == FSCK_FIX_yes ||
 			   (c->opts.nochanges &&
 			    !(flags & FSCK_CAN_IGNORE))) {
-			if (print)
-				bch_err(c, "%s, fixing", buf);
-			fix = true;
+			prt_str(out, ", fixing");
+			ret = -BCH_ERR_fsck_fix;
 		} else {
-			if (print)
-				bch_err(c, "%s, not fixing", buf);
-			fix = false;
+			prt_str(out, ", not fixing");
 		}
 	} else if (flags & FSCK_NEED_FSCK) {
-		if (print)
-			bch_err(c, "%s (run fsck to correct)", buf);
+		prt_str(out, " (run fsck to correct)");
 	} else {
-		if (print)
-			bch_err(c, "%s (repair unimplemented)", buf);
+		prt_str(out, " (repair unimplemented)");
+	}
+
+	if (ret == -BCH_ERR_fsck_ignore &&
+	    (c->opts.fix_errors == FSCK_FIX_exit ||
+	     !(flags & FSCK_CAN_IGNORE)))
+		ret = -BCH_ERR_fsck_errors_not_fixed;
+
+	if (print) {
+		if (bch2_fs_stdio_redirect(c))
+			bch2_print(c, "%s\n", out->buf);
+		else
+			bch2_print_string_as_lines(KERN_ERR, out->buf);
 	}
 
-	if (suppressing)
+	if (test_bit(BCH_FS_fsck_running, &c->flags) &&
+	    (ret != -BCH_ERR_fsck_fix &&
+	     ret != -BCH_ERR_fsck_ignore))
+		bch_err(c, "Unable to continue, halting");
+	else if (suppressing)
 		bch_err(c, "Ratelimiting new instances of previous error");
 
-	mutex_unlock(&c->fsck_error_lock);
+	if (s)
+		s->ret = ret;
+
+	mutex_unlock(&c->fsck_error_msgs_lock);
+
+	printbuf_exit(&buf);
 
-	if (fix)
-		set_bit(BCH_FS_FSCK_FIXED_ERRORS, &c->flags);
+	if (inconsistent)
+		bch2_inconsistent_error(c);
+
+	if (ret == -BCH_ERR_fsck_fix) {
+		set_bit(BCH_FS_errors_fixed, &c->flags);
+	} else {
+		set_bit(BCH_FS_errors_not_fixed, &c->flags);
+		set_bit(BCH_FS_error, &c->flags);
+	}
 
-	return fix				? FSCK_ERR_FIX
-		: flags & FSCK_CAN_IGNORE	? FSCK_ERR_IGNORE
-						: FSCK_ERR_EXIT;
+	return ret;
 }
 
 void bch2_flush_fsck_errs(struct bch_fs *c)
 {
 	struct fsck_err_state *s, *n;
 
-	mutex_lock(&c->fsck_error_lock);
-	set_bit(BCH_FS_FSCK_DONE, &c->flags);
+	mutex_lock(&c->fsck_error_msgs_lock);
 
-	list_for_each_entry_safe(s, n, &c->fsck_errors, list) {
-		if (s->nr > 10)
-			bch_err(c, "Saw %llu errors like:\n    %s", s->nr, s->buf);
+	list_for_each_entry_safe(s, n, &c->fsck_error_msgs, list) {
+		if (s->ratelimited && s->last_msg)
+			bch_err(c, "Saw %llu errors like:\n    %s", s->nr, s->last_msg);
 
 		list_del(&s->list);
+		kfree(s->last_msg);
 		kfree(s);
 	}
 
-	mutex_unlock(&c->fsck_error_lock);
+	mutex_unlock(&c->fsck_error_msgs_lock);
 }