[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#1033976: [pre-approve] unblock: golang-1.19/1.19.8-2



On 2023-04-05 19:10:42 +0800, Shengjing Zhu wrote:
> Package: release.debian.org
> Severity: normal
> User: release.debian.org@packages.debian.org
> Usertags: unblock
> X-Debbugs-Cc: golang-1.19@packages.debian.org, zhsj@debian.org
> Control: affects -1 + src:golang-1.19
> 
> Please unblock package golang-1.19

Please go ahead with upload to unstable

Cheers

> 
> [ Reason ]
> Two upstream minor versions, fixing 5 CVEs
> 
> + CVE-2023-24532: crypto/elliptic: incorrect P-256 ScalarMult and
>   ScalarBaseMult results
> + CVE-2023-24537: go/parser: infinite loop in parsing
> + CVE-2023-24538: html/template: backticks not treated as string delimiters
> + CVE-2023-24534: net/http, net/textproto: denial of service from excessive
>   memory allocation
> + CVE-2023-24536: net/http, net/textproto, mime/multipart: denial of
>   service from excessive resource consumption
> 
> [ Impact ]
> 
> Several security issues in the Go standard libraries.
> 
> [ Tests ]
> Besise the unittests upstream added in the new release, I have use the new
> version to build some Go packages. And the result is good.
> 
> [ Risks ]
> Toolchain package and no autopkgtest.
> 
> [ Checklist ]
>   [x] all changes are documented in the d/changelog
>   [x] I reviewed all changes and I approve them
>   [x] attach debdiff against the package in testing
> 
>       I attached the debdiff with
>       filterdiff --exclude '*_test.go' --exclude '*_windows*' --exclude '*/testdata/*' \
>         --exclude '*/go.mod' --exclude '*/go.sum' --exclude '*/modules.txt'
> 
> [ Other info ]
> It may be the last golang-1.19 version to be uploaded during freeze.
> The next release is expected (if no urgent CVE happens) to be May, which
> is probably hard-freeze time.
> 
> unblock golang-1.19/1.19.8-2

> diff -Nru golang-1.19-1.19.6/debian/changelog golang-1.19-1.19.8/debian/changelog
> --- golang-1.19-1.19.6/debian/changelog	2023-02-17 17:56:44.000000000 +0800
> +++ golang-1.19-1.19.8/debian/changelog	2023-04-05 02:15:56.000000000 +0800
> @@ -1,3 +1,25 @@
> +golang-1.19 (1.19.8-1) experimental; urgency=medium
> +
> +  * Team upload
> +  * New upstream version 1.19.8
> +    + CVE-2023-24537: go/parser: infinite loop in parsing
> +    + CVE-2023-24538: html/template: backticks not treated as string delimiters
> +    + CVE-2023-24534: net/http, net/textproto: denial of service from excessive
> +      memory allocation
> +    + CVE-2023-24536: net/http, net/textproto, mime/multipart: denial of
> +      service from excessive resource consumption
> +
> + -- Shengjing Zhu <zhsj@debian.org>  Wed, 05 Apr 2023 02:15:56 +0800
> +
> +golang-1.19 (1.19.7-1) experimental; urgency=medium
> +
> +  * Team upload
> +  * New upstream version 1.19.7
> +    + CVE-2023-24532: crypto/elliptic: incorrect P-256 ScalarMult and
> +      ScalarBaseMult results
> +
> + -- Shengjing Zhu <zhsj@debian.org>  Wed, 08 Mar 2023 13:54:08 +0800
> +
>  golang-1.19 (1.19.6-2) unstable; urgency=medium
>  
>    * Team upload
> diff -Nru golang-1.19-1.19.6/src/cmd/go/internal/work/exec.go golang-1.19-1.19.8/src/cmd/go/internal/work/exec.go
> --- golang-1.19-1.19.6/src/cmd/go/internal/work/exec.go	2023-02-14 01:38:43.000000000 +0800
> +++ golang-1.19-1.19.8/src/cmd/go/internal/work/exec.go	2023-03-30 05:15:17.000000000 +0800
> @@ -2764,6 +2764,36 @@
>  		}
>  	}
>  
> +	// Scrutinize CFLAGS and related for flags that might cause
> +	// problems if we are using internal linking (for example, use of
> +	// plugins, LTO, etc) by calling a helper routine that builds on
> +	// the existing CGO flags allow-lists. If we see anything
> +	// suspicious, emit a special token file "preferlinkext" (known to
> +	// the linker) in the object file to signal the that it should not
> +	// try to link internally and should revert to external linking.
> +	// The token we pass is a suggestion, not a mandate; if a user is
> +	// explicitly asking for a specific linkmode via the "-linkmode"
> +	// flag, the token will be ignored. NB: in theory we could ditch
> +	// the token approach and just pass a flag to the linker when we
> +	// eventually invoke it, and the linker flag could then be
> +	// documented (although coming up with a simple explanation of the
> +	// flag might be challenging). For more context see issues #58619,
> +	// #58620, and #58848.
> +	flagSources := []string{"CGO_CFLAGS", "CGO_CXXFLAGS", "CGO_FFLAGS"}
> +	flagLists := [][]string{cgoCFLAGS, cgoCXXFLAGS, cgoFFLAGS}
> +	if flagsNotCompatibleWithInternalLinking(flagSources, flagLists) {
> +		tokenFile := objdir + "preferlinkext"
> +		if cfg.BuildN || cfg.BuildX {
> +			b.Showcmd("", "echo > %s", tokenFile)
> +		}
> +		if !cfg.BuildN {
> +			if err := os.WriteFile(tokenFile, nil, 0666); err != nil {
> +				return nil, nil, err
> +			}
> +		}
> +		outObj = append(outObj, tokenFile)
> +	}
> +
>  	if cfg.BuildMSan {
>  		cgoCFLAGS = append([]string{"-fsanitize=memory"}, cgoCFLAGS...)
>  		cgoLDFLAGS = append([]string{"-fsanitize=memory"}, cgoLDFLAGS...)
> @@ -3012,6 +3042,24 @@
>  	return outGo, outObj, nil
>  }
>  
> +// flagsNotCompatibleWithInternalLinking scans the list of cgo
> +// compiler flags (C/C++/Fortran) looking for flags that might cause
> +// problems if the build in question uses internal linking. The
> +// primary culprits are use of plugins or use of LTO, but we err on
> +// the side of caution, supporting only those flags that are on the
> +// allow-list for safe flags from security perspective. Return is TRUE
> +// if a sensitive flag is found, FALSE otherwise.
> +func flagsNotCompatibleWithInternalLinking(sourceList []string, flagListList [][]string) bool {
> +	for i := range sourceList {
> +		sn := sourceList[i]
> +		fll := flagListList[i]
> +		if err := checkCompilerFlagsForInternalLink(sn, sn, fll); err != nil {
> +			return true
> +		}
> +	}
> +	return false
> +}
> +
>  // dynimport creates a Go source file named importGo containing
>  // //go:cgo_import_dynamic directives for each symbol or library
>  // dynamically imported by the object files outObj.
> diff -Nru golang-1.19-1.19.6/src/cmd/go/internal/work/security.go golang-1.19-1.19.8/src/cmd/go/internal/work/security.go
> --- golang-1.19-1.19.6/src/cmd/go/internal/work/security.go	2023-02-14 01:38:43.000000000 +0800
> +++ golang-1.19-1.19.8/src/cmd/go/internal/work/security.go	2023-03-30 05:15:17.000000000 +0800
> @@ -230,32 +230,55 @@
>  }
>  
>  func checkCompilerFlags(name, source string, list []string) error {
> -	return checkFlags(name, source, list, validCompilerFlags, validCompilerFlagsWithNextArg)
> +	checkOverrides := true
> +	return checkFlags(name, source, list, validCompilerFlags, validCompilerFlagsWithNextArg, checkOverrides)
>  }
>  
>  func checkLinkerFlags(name, source string, list []string) error {
> -	return checkFlags(name, source, list, validLinkerFlags, validLinkerFlagsWithNextArg)
> +	checkOverrides := true
> +	return checkFlags(name, source, list, validLinkerFlags, validLinkerFlagsWithNextArg, checkOverrides)
>  }
>  
> -func checkFlags(name, source string, list []string, valid []*lazyregexp.Regexp, validNext []string) error {
> +// checkCompilerFlagsForInternalLink returns an error if 'list'
> +// contains a flag or flags that may not be fully supported by
> +// internal linking (meaning that we should punt the link to the
> +// external linker).
> +func checkCompilerFlagsForInternalLink(name, source string, list []string) error {
> +	checkOverrides := false
> +	if err := checkFlags(name, source, list, validCompilerFlags, validCompilerFlagsWithNextArg, checkOverrides); err != nil {
> +		return err
> +	}
> +	// Currently the only flag on the allow list that causes problems
> +	// for the linker is "-flto"; check for it manually here.
> +	for _, fl := range list {
> +		if strings.HasPrefix(fl, "-flto") {
> +			return fmt.Errorf("flag %q triggers external linking", fl)
> +		}
> +	}
> +	return nil
> +}
> +
> +func checkFlags(name, source string, list []string, valid []*lazyregexp.Regexp, validNext []string, checkOverrides bool) error {
>  	// Let users override rules with $CGO_CFLAGS_ALLOW, $CGO_CFLAGS_DISALLOW, etc.
>  	var (
>  		allow    *regexp.Regexp
>  		disallow *regexp.Regexp
>  	)
> -	if env := cfg.Getenv("CGO_" + name + "_ALLOW"); env != "" {
> -		r, err := regexp.Compile(env)
> -		if err != nil {
> -			return fmt.Errorf("parsing $CGO_%s_ALLOW: %v", name, err)
> +	if checkOverrides {
> +		if env := cfg.Getenv("CGO_" + name + "_ALLOW"); env != "" {
> +			r, err := regexp.Compile(env)
> +			if err != nil {
> +				return fmt.Errorf("parsing $CGO_%s_ALLOW: %v", name, err)
> +			}
> +			allow = r
>  		}
> -		allow = r
> -	}
> -	if env := cfg.Getenv("CGO_" + name + "_DISALLOW"); env != "" {
> -		r, err := regexp.Compile(env)
> -		if err != nil {
> -			return fmt.Errorf("parsing $CGO_%s_DISALLOW: %v", name, err)
> +		if env := cfg.Getenv("CGO_" + name + "_DISALLOW"); env != "" {
> +			r, err := regexp.Compile(env)
> +			if err != nil {
> +				return fmt.Errorf("parsing $CGO_%s_DISALLOW: %v", name, err)
> +			}
> +			disallow = r
>  		}
> -		disallow = r
>  	}
>  
>  Args:
> diff -Nru golang-1.19-1.19.6/src/cmd/internal/obj/ppc64/asm9.go golang-1.19-1.19.8/src/cmd/internal/obj/ppc64/asm9.go
> --- golang-1.19-1.19.6/src/cmd/internal/obj/ppc64/asm9.go	2023-02-14 01:38:43.000000000 +0800
> +++ golang-1.19-1.19.8/src/cmd/internal/obj/ppc64/asm9.go	2023-03-30 05:15:17.000000000 +0800
> @@ -3136,8 +3136,13 @@
>  		if r == 0 {
>  			r = c.getimpliedreg(&p.From, p)
>  		}
> -		o1 = AOP_IRR(OP_ADDIS, uint32(p.To.Reg), uint32(r), uint32(high16adjusted(v)))
> -		o2 = AOP_IRR(c.opload(p.As), uint32(p.To.Reg), uint32(p.To.Reg), uint32(v))
> +		if o.a6 == C_REG {
> +			o1 = AOP_IRR(OP_ADDIS, uint32(p.To.Reg), uint32(r), uint32(high16adjusted(v)))
> +			o2 = AOP_IRR(c.opload(p.As), uint32(p.To.Reg), uint32(p.To.Reg), uint32(v))
> +		} else {
> +			o1 = AOP_IRR(OP_ADDIS, uint32(REGTMP), uint32(r), uint32(high16adjusted(v)))
> +			o2 = AOP_IRR(c.opload(p.As), uint32(p.To.Reg), uint32(REGTMP), uint32(v))
> +		}
>  
>  		// Sign extend MOVB if needed
>  		o3 = LOP_RRR(OP_EXTSB, uint32(p.To.Reg), uint32(p.To.Reg), 0)
> @@ -3516,8 +3521,8 @@
>  				rel.Type = objabi.R_ADDRPOWER_TOCREL_DS
>  			}
>  		default:
> -			reuseBaseReg := p.As != AFMOVD && p.As != AFMOVS
> -			// Reuse To.Reg as base register if not FP move.
> +			reuseBaseReg := o.a6 == C_REG
> +			// Reuse To.Reg as base register if it is a GPR.
>  			o1, o2 = c.symbolAccess(p.From.Sym, v, p.To.Reg, inst, reuseBaseReg)
>  		}
>  
> diff -Nru golang-1.19-1.19.6/src/cmd/link/internal/arm/asm.go golang-1.19-1.19.8/src/cmd/link/internal/arm/asm.go
> --- golang-1.19-1.19.6/src/cmd/link/internal/arm/asm.go	2023-02-14 01:38:43.000000000 +0800
> +++ golang-1.19-1.19.8/src/cmd/link/internal/arm/asm.go	2023-03-30 05:15:17.000000000 +0800
> @@ -396,11 +396,28 @@
>  		// laid out. Conservatively use a trampoline. This should be rare, as we lay out packages
>  		// in dependency order.
>  		if ldr.SymValue(rs) != 0 {
> -			// r.Add is the instruction
> -			// low 24-bit encodes the target address
> -			t = (ldr.SymValue(rs) + int64(signext24(r.Add()&0xffffff)*4) - (ldr.SymValue(s) + int64(r.Off()))) / 4
> +			// Workaround for issue #58425: it appears that the
> +			// external linker doesn't always take into account the
> +			// relocation addend when doing reachability checks. This
> +			// means that if you have a call from function XYZ at
> +			// offset 8 to runtime.duffzero with addend 800 (for
> +			// example), where the distance between the start of XYZ
> +			// and the start of runtime.duffzero is just over the
> +			// limit (by 100 bytes, say), you can get "relocation
> +			// doesn't fit" errors from the external linker. To deal
> +			// with this, ignore the addend when performing the
> +			// distance calculation (this assumes that we're only
> +			// handling backward jumps; ideally we might want to check
> +			// both with and without the addend).
> +			if ctxt.IsExternal() {
> +				t = (ldr.SymValue(rs) - (ldr.SymValue(s) + int64(r.Off()))) / 4
> +			} else {
> +				// r.Add is the instruction
> +				// low 24-bit encodes the target address
> +				t = (ldr.SymValue(rs) + int64(signext24(r.Add()&0xffffff)*4) - (ldr.SymValue(s) + int64(r.Off()))) / 4
> +			}
>  		}
> -		if t > 0x7fffff || t < -0x800000 || ldr.SymValue(rs) == 0 || (*ld.FlagDebugTramp > 1 && ldr.SymPkg(s) != ldr.SymPkg(rs)) {
> +		if t > 0x7fffff || t <= -0x800000 || ldr.SymValue(rs) == 0 || (*ld.FlagDebugTramp > 1 && ldr.SymPkg(s) != ldr.SymPkg(rs)) {
>  			// direct call too far, need to insert trampoline.
>  			// look up existing trampolines first. if we found one within the range
>  			// of direct call, we can reuse it. otherwise create a new one.
> diff -Nru golang-1.19-1.19.6/src/cmd/link/internal/ld/config.go golang-1.19-1.19.8/src/cmd/link/internal/ld/config.go
> --- golang-1.19-1.19.6/src/cmd/link/internal/ld/config.go	2023-02-14 01:38:43.000000000 +0800
> +++ golang-1.19-1.19.8/src/cmd/link/internal/ld/config.go	2023-03-30 05:15:17.000000000 +0800
> @@ -280,7 +280,11 @@
>  			ctxt.LinkMode = LinkExternal
>  			via = "via GO_EXTLINK_ENABLED "
>  		default:
> -			if extNeeded || (iscgo && externalobj) {
> +			preferExternal := len(preferlinkext) != 0
> +			if preferExternal && ctxt.Debugvlog > 0 {
> +				ctxt.Logf("external linking prefer list is %v\n", preferlinkext)
> +			}
> +			if extNeeded || (iscgo && (externalobj || preferExternal)) {
>  				ctxt.LinkMode = LinkExternal
>  			} else {
>  				ctxt.LinkMode = LinkInternal
> diff -Nru golang-1.19-1.19.6/src/cmd/link/internal/ld/lib.go golang-1.19-1.19.8/src/cmd/link/internal/ld/lib.go
> --- golang-1.19-1.19.6/src/cmd/link/internal/ld/lib.go	2023-02-14 01:38:43.000000000 +0800
> +++ golang-1.19-1.19.8/src/cmd/link/internal/ld/lib.go	2023-03-30 05:15:17.000000000 +0800
> @@ -349,6 +349,12 @@
>  	// any of these objects, we must link externally. Issue 52863.
>  	dynimportfail []string
>  
> +	// preferlinkext is a list of packages for which the Go command
> +	// noticed use of peculiar C flags. If we see any of these,
> +	// default to linking externally unless overridden by the
> +	// user. See issues #58619, #58620, and #58848.
> +	preferlinkext []string
> +
>  	// unknownObjFormat is set to true if we see an object whose
>  	// format we don't recognize.
>  	unknownObjFormat = false
> @@ -1043,6 +1049,13 @@
>  		if arhdr.name == "dynimportfail" {
>  			dynimportfail = append(dynimportfail, lib.Pkg)
>  		}
> +		if arhdr.name == "preferlinkext" {
> +			// Ignore this directive if -linkmode has been
> +			// set explicitly.
> +			if ctxt.LinkMode == LinkAuto {
> +				preferlinkext = append(preferlinkext, lib.Pkg)
> +			}
> +		}
>  
>  		// Skip other special (non-object-file) sections that
>  		// build tools may have added. Such sections must have
> diff -Nru golang-1.19-1.19.6/src/crypto/internal/nistec/p256_asm.go golang-1.19-1.19.8/src/crypto/internal/nistec/p256_asm.go
> --- golang-1.19-1.19.6/src/crypto/internal/nistec/p256_asm.go	2023-02-14 01:38:45.000000000 +0800
> +++ golang-1.19-1.19.8/src/crypto/internal/nistec/p256_asm.go	2023-03-30 05:15:20.000000000 +0800
> @@ -365,6 +365,21 @@
>  // Montgomery domain (with R 2²⁵⁶) as four uint64 limbs in little-endian order.
>  type p256OrdElement [4]uint64
>  
> +// p256OrdReduce ensures s is in the range [0, ord(G)-1].
> +func p256OrdReduce(s *p256OrdElement) {
> +	// Since 2 * ord(G) > 2²⁵⁶, we can just conditionally subtract ord(G),
> +	// keeping the result if it doesn't underflow.
> +	t0, b := bits.Sub64(s[0], 0xf3b9cac2fc632551, 0)
> +	t1, b := bits.Sub64(s[1], 0xbce6faada7179e84, b)
> +	t2, b := bits.Sub64(s[2], 0xffffffffffffffff, b)
> +	t3, b := bits.Sub64(s[3], 0xffffffff00000000, b)
> +	tMask := b - 1 // zero if subtraction underflowed
> +	s[0] ^= (t0 ^ s[0]) & tMask
> +	s[1] ^= (t1 ^ s[1]) & tMask
> +	s[2] ^= (t2 ^ s[2]) & tMask
> +	s[3] ^= (t3 ^ s[3]) & tMask
> +}
> +
>  // Add sets q = p1 + p2, and returns q. The points may overlap.
>  func (q *P256Point) Add(r1, r2 *P256Point) *P256Point {
>  	var sum, double P256Point
> @@ -394,6 +409,7 @@
>  	}
>  	scalarReversed := new(p256OrdElement)
>  	p256OrdBigToLittle(scalarReversed, (*[32]byte)(scalar))
> +	p256OrdReduce(scalarReversed)
>  
>  	r.p256BaseMult(scalarReversed)
>  	return r, nil
> @@ -408,6 +424,7 @@
>  	}
>  	scalarReversed := new(p256OrdElement)
>  	p256OrdBigToLittle(scalarReversed, (*[32]byte)(scalar))
> +	p256OrdReduce(scalarReversed)
>  
>  	r.Set(q).p256ScalarMult(scalarReversed)
>  	return r, nil
> diff -Nru golang-1.19-1.19.6/src/crypto/internal/nistec/p256_asm_ordinv.go golang-1.19-1.19.8/src/crypto/internal/nistec/p256_asm_ordinv.go
> --- golang-1.19-1.19.6/src/crypto/internal/nistec/p256_asm_ordinv.go	2023-02-14 01:38:45.000000000 +0800
> +++ golang-1.19-1.19.8/src/crypto/internal/nistec/p256_asm_ordinv.go	2023-03-30 05:15:20.000000000 +0800
> @@ -25,6 +25,7 @@
>  
>  	x := new(p256OrdElement)
>  	p256OrdBigToLittle(x, (*[32]byte)(k))
> +	p256OrdReduce(x)
>  
>  	// Inversion is implemented as exponentiation by n - 2, per Fermat's little theorem.
>  	//
> diff -Nru golang-1.19-1.19.6/src/crypto/x509/root_darwin.go golang-1.19-1.19.8/src/crypto/x509/root_darwin.go
> --- golang-1.19-1.19.6/src/crypto/x509/root_darwin.go	2023-02-14 01:38:45.000000000 +0800
> +++ golang-1.19-1.19.8/src/crypto/x509/root_darwin.go	2023-03-30 05:15:20.000000000 +0800
> @@ -25,9 +25,10 @@
>  				return nil, err
>  			}
>  			sc, err := macOS.SecCertificateCreateWithData(c.Raw)
> -			if err == nil {
> -				macOS.CFArrayAppendValue(certs, sc)
> +			if err != nil {
> +				return nil, err
>  			}
> +			macOS.CFArrayAppendValue(certs, sc)
>  		}
>  	}
>  
> diff -Nru golang-1.19-1.19.6/src/go/scanner/scanner.go golang-1.19-1.19.8/src/go/scanner/scanner.go
> --- golang-1.19-1.19.6/src/go/scanner/scanner.go	2023-02-14 01:38:46.000000000 +0800
> +++ golang-1.19-1.19.8/src/go/scanner/scanner.go	2023-03-30 05:15:20.000000000 +0800
> @@ -246,13 +246,16 @@
>  		return
>  	}
>  
> +	// Put a cap on the maximum size of line and column numbers.
> +	// 30 bits allows for some additional space before wrapping an int32.
> +	const maxLineCol = 1<<30 - 1
>  	var line, col int
>  	i2, n2, ok2 := trailingDigits(text[:i-1])
>  	if ok2 {
>  		//line filename:line:col
>  		i, i2 = i2, i
>  		line, col = n2, n
> -		if col == 0 {
> +		if col == 0 || col > maxLineCol {
>  			s.error(offs+i2, "invalid column number: "+string(text[i2:]))
>  			return
>  		}
> @@ -262,7 +265,7 @@
>  		line = n
>  	}
>  
> -	if line == 0 {
> +	if line == 0 || line > maxLineCol {
>  		s.error(offs+i, "invalid line number: "+string(text[i:]))
>  		return
>  	}
> diff -Nru golang-1.19-1.19.6/src/html/template/context.go golang-1.19-1.19.8/src/html/template/context.go
> --- golang-1.19-1.19.6/src/html/template/context.go	2023-02-14 01:38:46.000000000 +0800
> +++ golang-1.19-1.19.8/src/html/template/context.go	2023-03-30 05:15:20.000000000 +0800
> @@ -120,6 +120,8 @@
>  	stateJSDqStr
>  	// stateJSSqStr occurs inside a JavaScript single quoted string.
>  	stateJSSqStr
> +	// stateJSBqStr occurs inside a JavaScript back quoted string.
> +	stateJSBqStr
>  	// stateJSRegexp occurs inside a JavaScript regexp literal.
>  	stateJSRegexp
>  	// stateJSBlockCmt occurs inside a JavaScript /* block comment */.
> diff -Nru golang-1.19-1.19.6/src/html/template/error.go golang-1.19-1.19.8/src/html/template/error.go
> --- golang-1.19-1.19.6/src/html/template/error.go	2023-02-14 01:38:46.000000000 +0800
> +++ golang-1.19-1.19.8/src/html/template/error.go	2023-03-30 05:15:20.000000000 +0800
> @@ -214,6 +214,19 @@
>  	//   pipeline occurs in an unquoted attribute value context, "html" is
>  	//   disallowed. Avoid using "html" and "urlquery" entirely in new templates.
>  	ErrPredefinedEscaper
> +
> +	// errJSTmplLit: "... appears in a JS template literal"
> +	// Example:
> +	//     <script>var tmpl = `{{.Interp}`</script>
> +	// Discussion:
> +	//   Package html/template does not support actions inside of JS template
> +	//   literals.
> +	//
> +	// TODO(rolandshoemaker): we cannot add this as an exported error in a minor
> +	// release, since it is backwards incompatible with the other minor
> +	// releases. As such we need to leave it unexported, and then we'll add it
> +	// in the next major release.
> +	errJSTmplLit
>  )
>  
>  func (e *Error) Error() string {
> diff -Nru golang-1.19-1.19.6/src/html/template/escape.go golang-1.19-1.19.8/src/html/template/escape.go
> --- golang-1.19-1.19.6/src/html/template/escape.go	2023-02-14 01:38:46.000000000 +0800
> +++ golang-1.19-1.19.8/src/html/template/escape.go	2023-03-30 05:15:20.000000000 +0800
> @@ -8,6 +8,7 @@
>  	"bytes"
>  	"fmt"
>  	"html"
> +	"internal/godebug"
>  	"io"
>  	"text/template"
>  	"text/template/parse"
> @@ -223,6 +224,16 @@
>  		c.jsCtx = jsCtxDivOp
>  	case stateJSDqStr, stateJSSqStr:
>  		s = append(s, "_html_template_jsstrescaper")
> +	case stateJSBqStr:
> +		debugAllowActionJSTmpl := godebug.Get("jstmpllitinterp")
> +		if debugAllowActionJSTmpl == "1" {
> +			s = append(s, "_html_template_jsstrescaper")
> +		} else {
> +			return context{
> +				state: stateError,
> +				err:   errorf(errJSTmplLit, n, n.Line, "%s appears in a JS template literal", n),
> +			}
> +		}
>  	case stateJSRegexp:
>  		s = append(s, "_html_template_jsregexpescaper")
>  	case stateCSS:
> diff -Nru golang-1.19-1.19.6/src/html/template/jsctx_string.go golang-1.19-1.19.8/src/html/template/jsctx_string.go
> --- golang-1.19-1.19.6/src/html/template/jsctx_string.go	2023-02-14 01:38:46.000000000 +0800
> +++ golang-1.19-1.19.8/src/html/template/jsctx_string.go	2023-03-30 05:15:21.000000000 +0800
> @@ -4,6 +4,15 @@
>  
>  import "strconv"
>  
> +func _() {
> +	// An "invalid array index" compiler error signifies that the constant values have changed.
> +	// Re-run the stringer command to generate them again.
> +	var x [1]struct{}
> +	_ = x[jsCtxRegexp-0]
> +	_ = x[jsCtxDivOp-1]
> +	_ = x[jsCtxUnknown-2]
> +}
> +
>  const _jsCtx_name = "jsCtxRegexpjsCtxDivOpjsCtxUnknown"
>  
>  var _jsCtx_index = [...]uint8{0, 11, 21, 33}
> diff -Nru golang-1.19-1.19.6/src/html/template/js.go golang-1.19-1.19.8/src/html/template/js.go
> --- golang-1.19-1.19.6/src/html/template/js.go	2023-02-14 01:38:46.000000000 +0800
> +++ golang-1.19-1.19.8/src/html/template/js.go	2023-03-30 05:15:21.000000000 +0800
> @@ -308,6 +308,7 @@
>  	// Encode HTML specials as hex so the output can be embedded
>  	// in HTML attributes without further encoding.
>  	'"':  `\u0022`,
> +	'`':  `\u0060`,
>  	'&':  `\u0026`,
>  	'\'': `\u0027`,
>  	'+':  `\u002b`,
> @@ -331,6 +332,7 @@
>  	'"':  `\u0022`,
>  	'&':  `\u0026`,
>  	'\'': `\u0027`,
> +	'`':  `\u0060`,
>  	'+':  `\u002b`,
>  	'/':  `\/`,
>  	'<':  `\u003c`,
> diff -Nru golang-1.19-1.19.6/src/html/template/state_string.go golang-1.19-1.19.8/src/html/template/state_string.go
> --- golang-1.19-1.19.6/src/html/template/state_string.go	2023-02-14 01:38:46.000000000 +0800
> +++ golang-1.19-1.19.8/src/html/template/state_string.go	2023-03-30 05:15:21.000000000 +0800
> @@ -4,9 +4,42 @@
>  
>  import "strconv"
>  
> -const _state_name = "stateTextstateTagstateAttrNamestateAfterNamestateBeforeValuestateHTMLCmtstateRCDATAstateAttrstateURLstateSrcsetstateJSstateJSDqStrstateJSSqStrstateJSRegexpstateJSBlockCmtstateJSLineCmtstateCSSstateCSSDqStrstateCSSSqStrstateCSSDqURLstateCSSSqURLstateCSSURLstateCSSBlockCmtstateCSSLineCmtstateError"
> +func _() {
> +	// An "invalid array index" compiler error signifies that the constant values have changed.
> +	// Re-run the stringer command to generate them again.
> +	var x [1]struct{}
> +	_ = x[stateText-0]
> +	_ = x[stateTag-1]
> +	_ = x[stateAttrName-2]
> +	_ = x[stateAfterName-3]
> +	_ = x[stateBeforeValue-4]
> +	_ = x[stateHTMLCmt-5]
> +	_ = x[stateRCDATA-6]
> +	_ = x[stateAttr-7]
> +	_ = x[stateURL-8]
> +	_ = x[stateSrcset-9]
> +	_ = x[stateJS-10]
> +	_ = x[stateJSDqStr-11]
> +	_ = x[stateJSSqStr-12]
> +	_ = x[stateJSBqStr-13]
> +	_ = x[stateJSRegexp-14]
> +	_ = x[stateJSBlockCmt-15]
> +	_ = x[stateJSLineCmt-16]
> +	_ = x[stateCSS-17]
> +	_ = x[stateCSSDqStr-18]
> +	_ = x[stateCSSSqStr-19]
> +	_ = x[stateCSSDqURL-20]
> +	_ = x[stateCSSSqURL-21]
> +	_ = x[stateCSSURL-22]
> +	_ = x[stateCSSBlockCmt-23]
> +	_ = x[stateCSSLineCmt-24]
> +	_ = x[stateError-25]
> +	_ = x[stateDead-26]
> +}
>  
> -var _state_index = [...]uint16{0, 9, 17, 30, 44, 60, 72, 83, 92, 100, 111, 118, 130, 142, 155, 170, 184, 192, 205, 218, 231, 244, 255, 271, 286, 296}
> +const _state_name = "stateTextstateTagstateAttrNamestateAfterNamestateBeforeValuestateHTMLCmtstateRCDATAstateAttrstateURLstateSrcsetstateJSstateJSDqStrstateJSSqStrstateJSBqStrstateJSRegexpstateJSBlockCmtstateJSLineCmtstateCSSstateCSSDqStrstateCSSSqStrstateCSSDqURLstateCSSSqURLstateCSSURLstateCSSBlockCmtstateCSSLineCmtstateErrorstateDead"
> +
> +var _state_index = [...]uint16{0, 9, 17, 30, 44, 60, 72, 83, 92, 100, 111, 118, 130, 142, 154, 167, 182, 196, 204, 217, 230, 243, 256, 267, 283, 298, 308, 317}
>  
>  func (i state) String() string {
>  	if i >= state(len(_state_index)-1) {
> diff -Nru golang-1.19-1.19.6/src/html/template/transition.go golang-1.19-1.19.8/src/html/template/transition.go
> --- golang-1.19-1.19.6/src/html/template/transition.go	2023-02-14 01:38:46.000000000 +0800
> +++ golang-1.19-1.19.8/src/html/template/transition.go	2023-03-30 05:15:21.000000000 +0800
> @@ -27,6 +27,7 @@
>  	stateJS:          tJS,
>  	stateJSDqStr:     tJSDelimited,
>  	stateJSSqStr:     tJSDelimited,
> +	stateJSBqStr:     tJSDelimited,
>  	stateJSRegexp:    tJSDelimited,
>  	stateJSBlockCmt:  tBlockCmt,
>  	stateJSLineCmt:   tLineCmt,
> @@ -262,7 +263,7 @@
>  
>  // tJS is the context transition function for the JS state.
>  func tJS(c context, s []byte) (context, int) {
> -	i := bytes.IndexAny(s, `"'/`)
> +	i := bytes.IndexAny(s, "\"`'/")
>  	if i == -1 {
>  		// Entire input is non string, comment, regexp tokens.
>  		c.jsCtx = nextJSCtx(s, c.jsCtx)
> @@ -274,6 +275,8 @@
>  		c.state, c.jsCtx = stateJSDqStr, jsCtxRegexp
>  	case '\'':
>  		c.state, c.jsCtx = stateJSSqStr, jsCtxRegexp
> +	case '`':
> +		c.state, c.jsCtx = stateJSBqStr, jsCtxRegexp
>  	case '/':
>  		switch {
>  		case i+1 < len(s) && s[i+1] == '/':
> @@ -303,6 +306,8 @@
>  	switch c.state {
>  	case stateJSSqStr:
>  		specials = `\'`
> +	case stateJSBqStr:
> +		specials = "`\\"
>  	case stateJSRegexp:
>  		specials = `\/[]`
>  	}
> diff -Nru golang-1.19-1.19.6/src/internal/poll/fd_poll_runtime.go golang-1.19-1.19.8/src/internal/poll/fd_poll_runtime.go
> --- golang-1.19-1.19.6/src/internal/poll/fd_poll_runtime.go	2023-02-14 01:38:46.000000000 +0800
> +++ golang-1.19-1.19.8/src/internal/poll/fd_poll_runtime.go	2023-03-30 05:15:21.000000000 +0800
> @@ -23,7 +23,7 @@
>  func runtime_pollOpen(fd uintptr) (uintptr, int)
>  func runtime_pollClose(ctx uintptr)
>  func runtime_pollWait(ctx uintptr, mode int) int
> -func runtime_pollWaitCanceled(ctx uintptr, mode int) int
> +func runtime_pollWaitCanceled(ctx uintptr, mode int)
>  func runtime_pollReset(ctx uintptr, mode int) int
>  func runtime_pollSetDeadline(ctx uintptr, d int64, mode int)
>  func runtime_pollUnblock(ctx uintptr)
> diff -Nru golang-1.19-1.19.6/src/mime/multipart/formdata.go golang-1.19-1.19.8/src/mime/multipart/formdata.go
> --- golang-1.19-1.19.6/src/mime/multipart/formdata.go	2023-02-14 01:38:47.000000000 +0800
> +++ golang-1.19-1.19.8/src/mime/multipart/formdata.go	2023-03-30 05:15:21.000000000 +0800
> @@ -12,6 +12,7 @@
>  	"math"
>  	"net/textproto"
>  	"os"
> +	"strconv"
>  )
>  
>  // ErrMessageTooLarge is returned by ReadForm if the message form
> @@ -41,6 +42,15 @@
>  	numDiskFiles := 0
>  	multipartFiles := godebug.Get("multipartfiles")
>  	combineFiles := multipartFiles != "distinct"
> +	maxParts := 1000
> +	multipartMaxParts := godebug.Get("multipartmaxparts")
> +	if multipartMaxParts != "" {
> +		if v, err := strconv.Atoi(multipartMaxParts); err == nil && v >= 0 {
> +			maxParts = v
> +		}
> +	}
> +	maxHeaders := maxMIMEHeaders()
> +
>  	defer func() {
>  		if file != nil {
>  			if cerr := file.Close(); err == nil {
> @@ -84,14 +94,19 @@
>  			maxMemoryBytes = math.MaxInt64
>  		}
>  	}
> +	var copyBuf []byte
>  	for {
> -		p, err := r.nextPart(false, maxMemoryBytes)
> +		p, err := r.nextPart(false, maxMemoryBytes, maxHeaders)
>  		if err == io.EOF {
>  			break
>  		}
>  		if err != nil {
>  			return nil, err
>  		}
> +		if maxParts <= 0 {
> +			return nil, ErrMessageTooLarge
> +		}
> +		maxParts--
>  
>  		name := p.FormName()
>  		if name == "" {
> @@ -102,8 +117,9 @@
>  		// Multiple values for the same key (one map entry, longer slice) are cheaper
>  		// than the same number of values for different keys (many map entries), but
>  		// using a consistent per-value cost for overhead is simpler.
> +		const mapEntryOverhead = 200
>  		maxMemoryBytes -= int64(len(name))
> -		maxMemoryBytes -= 100 // map overhead
> +		maxMemoryBytes -= mapEntryOverhead
>  		if maxMemoryBytes < 0 {
>  			// We can't actually take this path, since nextPart would already have
>  			// rejected the MIME headers for being too large. Check anyway.
> @@ -127,10 +143,16 @@
>  		}
>  
>  		// file, store in memory or on disk
> +		const fileHeaderSize = 100
>  		maxMemoryBytes -= mimeHeaderSize(p.Header)
> +		maxMemoryBytes -= mapEntryOverhead
> +		maxMemoryBytes -= fileHeaderSize
>  		if maxMemoryBytes < 0 {
>  			return nil, ErrMessageTooLarge
>  		}
> +		for _, v := range p.Header {
> +			maxHeaders -= int64(len(v))
> +		}
>  		fh := &FileHeader{
>  			Filename: filename,
>  			Header:   p.Header,
> @@ -147,14 +169,22 @@
>  				}
>  			}
>  			numDiskFiles++
> -			size, err := io.Copy(file, io.MultiReader(&b, p))
> +			if _, err := file.Write(b.Bytes()); err != nil {
> +				return nil, err
> +			}
> +			if copyBuf == nil {
> +				copyBuf = make([]byte, 32*1024) // same buffer size as io.Copy uses
> +			}
> +			// os.File.ReadFrom will allocate its own copy buffer if we let io.Copy use it.
> +			type writerOnly struct{ io.Writer }
> +			remainingSize, err := io.CopyBuffer(writerOnly{file}, p, copyBuf)
>  			if err != nil {
>  				return nil, err
>  			}
>  			fh.tmpfile = file.Name()
> -			fh.Size = size
> +			fh.Size = int64(b.Len()) + remainingSize
>  			fh.tmpoff = fileOff
> -			fileOff += size
> +			fileOff += fh.Size
>  			if !combineFiles {
>  				if err := file.Close(); err != nil {
>  					return nil, err
> @@ -174,9 +204,10 @@
>  }
>  
>  func mimeHeaderSize(h textproto.MIMEHeader) (size int64) {
> +	size = 400
>  	for k, vs := range h {
>  		size += int64(len(k))
> -		size += 100 // map entry overhead
> +		size += 200 // map entry overhead
>  		for _, v := range vs {
>  			size += int64(len(v))
>  		}
> diff -Nru golang-1.19-1.19.6/src/mime/multipart/multipart.go golang-1.19-1.19.8/src/mime/multipart/multipart.go
> --- golang-1.19-1.19.6/src/mime/multipart/multipart.go	2023-02-14 01:38:47.000000000 +0800
> +++ golang-1.19-1.19.8/src/mime/multipart/multipart.go	2023-03-30 05:15:21.000000000 +0800
> @@ -16,11 +16,13 @@
>  	"bufio"
>  	"bytes"
>  	"fmt"
> +	"internal/godebug"
>  	"io"
>  	"mime"
>  	"mime/quotedprintable"
>  	"net/textproto"
>  	"path/filepath"
> +	"strconv"
>  	"strings"
>  )
>  
> @@ -128,12 +130,12 @@
>  	return n, r.err
>  }
>  
> -func newPart(mr *Reader, rawPart bool, maxMIMEHeaderSize int64) (*Part, error) {
> +func newPart(mr *Reader, rawPart bool, maxMIMEHeaderSize, maxMIMEHeaders int64) (*Part, error) {
>  	bp := &Part{
>  		Header: make(map[string][]string),
>  		mr:     mr,
>  	}
> -	if err := bp.populateHeaders(maxMIMEHeaderSize); err != nil {
> +	if err := bp.populateHeaders(maxMIMEHeaderSize, maxMIMEHeaders); err != nil {
>  		return nil, err
>  	}
>  	bp.r = partReader{bp}
> @@ -149,9 +151,9 @@
>  	return bp, nil
>  }
>  
> -func (p *Part) populateHeaders(maxMIMEHeaderSize int64) error {
> +func (p *Part) populateHeaders(maxMIMEHeaderSize, maxMIMEHeaders int64) error {
>  	r := textproto.NewReader(p.mr.bufReader)
> -	header, err := readMIMEHeader(r, maxMIMEHeaderSize)
> +	header, err := readMIMEHeader(r, maxMIMEHeaderSize, maxMIMEHeaders)
>  	if err == nil {
>  		p.Header = header
>  	}
> @@ -330,6 +332,19 @@
>  // including header keys, values, and map overhead.
>  const maxMIMEHeaderSize = 10 << 20
>  
> +func maxMIMEHeaders() int64 {
> +	// multipartMaxHeaders is the maximum number of header entries NextPart will return,
> +	// as well as the maximum combined total of header entries Reader.ReadForm will return
> +	// in FileHeaders.
> +	multipartMaxHeaders := godebug.Get("multipartmaxheaders")
> +	if multipartMaxHeaders != "" {
> +		if v, err := strconv.ParseInt(multipartMaxHeaders, 10, 64); err == nil && v >= 0 {
> +			return v
> +		}
> +	}
> +	return 10000
> +}
> +
>  // NextPart returns the next part in the multipart or an error.
>  // When there are no more parts, the error io.EOF is returned.
>  //
> @@ -337,7 +352,7 @@
>  // has a value of "quoted-printable", that header is instead
>  // hidden and the body is transparently decoded during Read calls.
>  func (r *Reader) NextPart() (*Part, error) {
> -	return r.nextPart(false, maxMIMEHeaderSize)
> +	return r.nextPart(false, maxMIMEHeaderSize, maxMIMEHeaders())
>  }
>  
>  // NextRawPart returns the next part in the multipart or an error.
> @@ -346,10 +361,10 @@
>  // Unlike NextPart, it does not have special handling for
>  // "Content-Transfer-Encoding: quoted-printable".
>  func (r *Reader) NextRawPart() (*Part, error) {
> -	return r.nextPart(true, maxMIMEHeaderSize)
> +	return r.nextPart(true, maxMIMEHeaderSize, maxMIMEHeaders())
>  }
>  
> -func (r *Reader) nextPart(rawPart bool, maxMIMEHeaderSize int64) (*Part, error) {
> +func (r *Reader) nextPart(rawPart bool, maxMIMEHeaderSize, maxMIMEHeaders int64) (*Part, error) {
>  	if r.currentPart != nil {
>  		r.currentPart.Close()
>  	}
> @@ -374,7 +389,7 @@
>  
>  		if r.isBoundaryDelimiterLine(line) {
>  			r.partsRead++
> -			bp, err := newPart(r, rawPart, maxMIMEHeaderSize)
> +			bp, err := newPart(r, rawPart, maxMIMEHeaderSize, maxMIMEHeaders)
>  			if err != nil {
>  				return nil, err
>  			}
> diff -Nru golang-1.19-1.19.6/src/mime/multipart/readmimeheader.go golang-1.19-1.19.8/src/mime/multipart/readmimeheader.go
> --- golang-1.19-1.19.6/src/mime/multipart/readmimeheader.go	2023-02-14 01:38:47.000000000 +0800
> +++ golang-1.19-1.19.8/src/mime/multipart/readmimeheader.go	2023-03-30 05:15:21.000000000 +0800
> @@ -11,4 +11,4 @@
>  // readMIMEHeader is defined in package net/textproto.
>  //
>  //go:linkname readMIMEHeader net/textproto.readMIMEHeader
> -func readMIMEHeader(r *textproto.Reader, lim int64) (textproto.MIMEHeader, error)
> +func readMIMEHeader(r *textproto.Reader, maxMemory, maxHeaders int64) (textproto.MIMEHeader, error)
> diff -Nru golang-1.19-1.19.6/src/net/textproto/reader.go golang-1.19-1.19.8/src/net/textproto/reader.go
> --- golang-1.19-1.19.6/src/net/textproto/reader.go	2023-02-14 01:38:47.000000000 +0800
> +++ golang-1.19-1.19.8/src/net/textproto/reader.go	2023-03-30 05:15:21.000000000 +0800
> @@ -483,23 +483,32 @@
>  //		"Long-Key": {"Even Longer Value"},
>  //	}
>  func (r *Reader) ReadMIMEHeader() (MIMEHeader, error) {
> -	return readMIMEHeader(r, math.MaxInt64)
> +	return readMIMEHeader(r, math.MaxInt64, math.MaxInt64)
>  }
>  
>  // readMIMEHeader is a version of ReadMIMEHeader which takes a limit on the header size.
>  // It is called by the mime/multipart package.
> -func readMIMEHeader(r *Reader, lim int64) (MIMEHeader, error) {
> +func readMIMEHeader(r *Reader, maxMemory, maxHeaders int64) (MIMEHeader, error) {
>  	// Avoid lots of small slice allocations later by allocating one
>  	// large one ahead of time which we'll cut up into smaller
>  	// slices. If this isn't big enough later, we allocate small ones.
>  	var strs []string
> -	hint := r.upcomingHeaderNewlines()
> +	hint := r.upcomingHeaderKeys()
>  	if hint > 0 {
> +		if hint > 1000 {
> +			hint = 1000 // set a cap to avoid overallocation
> +		}
>  		strs = make([]string, hint)
>  	}
>  
>  	m := make(MIMEHeader, hint)
>  
> +	// Account for 400 bytes of overhead for the MIMEHeader, plus 200 bytes per entry.
> +	// Benchmarking map creation as of go1.20, a one-entry MIMEHeader is 416 bytes and large
> +	// MIMEHeaders average about 200 bytes per entry.
> +	maxMemory -= 400
> +	const mapEntryOverhead = 200
> +
>  	// The first line cannot start with a leading space.
>  	if buf, err := r.R.Peek(1); err == nil && (buf[0] == ' ' || buf[0] == '\t') {
>  		line, err := r.readLineSlice()
> @@ -529,16 +538,21 @@
>  			continue
>  		}
>  
> +		maxHeaders--
> +		if maxHeaders < 0 {
> +			return nil, errors.New("message too large")
> +		}
> +
>  		// Skip initial spaces in value.
>  		value := string(bytes.TrimLeft(v, " \t"))
>  
>  		vv := m[key]
>  		if vv == nil {
> -			lim -= int64(len(key))
> -			lim -= 100 // map entry overhead
> +			maxMemory -= int64(len(key))
> +			maxMemory -= mapEntryOverhead
>  		}
> -		lim -= int64(len(value))
> -		if lim < 0 {
> +		maxMemory -= int64(len(value))
> +		if maxMemory < 0 {
>  			// TODO: This should be a distinguishable error (ErrMessageTooLarge)
>  			// to allow mime/multipart to detect it.
>  			return m, errors.New("message too large")
> @@ -577,9 +591,9 @@
>  
>  var nl = []byte("\n")
>  
> -// upcomingHeaderNewlines returns an approximation of the number of newlines
> +// upcomingHeaderKeys returns an approximation of the number of keys
>  // that will be in this header. If it gets confused, it returns 0.
> -func (r *Reader) upcomingHeaderNewlines() (n int) {
> +func (r *Reader) upcomingHeaderKeys() (n int) {
>  	// Try to determine the 'hint' size.
>  	r.R.Peek(1) // force a buffer load if empty
>  	s := r.R.Buffered()
> @@ -587,7 +601,20 @@
>  		return
>  	}
>  	peek, _ := r.R.Peek(s)
> -	return bytes.Count(peek, nl)
> +	for len(peek) > 0 && n < 1000 {
> +		var line []byte
> +		line, peek, _ = bytes.Cut(peek, nl)
> +		if len(line) == 0 || (len(line) == 1 && line[0] == '\r') {
> +			// Blank line separating headers from the body.
> +			break
> +		}
> +		if line[0] == ' ' || line[0] == '\t' {
> +			// Folded continuation of the previous line.
> +			continue
> +		}
> +		n++
> +	}
> +	return n
>  }
>  
>  // CanonicalMIMEHeaderKey returns the canonical format of the
> diff -Nru golang-1.19-1.19.6/src/os/signal/internal/pty/pty.go golang-1.19-1.19.8/src/os/signal/internal/pty/pty.go
> --- golang-1.19-1.19.6/src/os/signal/internal/pty/pty.go	2023-02-14 01:38:47.000000000 +0800
> +++ golang-1.19-1.19.8/src/os/signal/internal/pty/pty.go	2023-03-30 05:15:22.000000000 +0800
> @@ -42,14 +42,14 @@
>  // Open returns a control pty and the name of the linked process tty.
>  func Open() (pty *os.File, processTTY string, err error) {
>  	m, err := C.posix_openpt(C.O_RDWR)
> -	if err != nil {
> +	if m < 0 {
>  		return nil, "", ptyError("posix_openpt", err)
>  	}
> -	if _, err := C.grantpt(m); err != nil {
> +	if res, err := C.grantpt(m); res < 0 {
>  		C.close(m)
>  		return nil, "", ptyError("grantpt", err)
>  	}
> -	if _, err := C.unlockpt(m); err != nil {
> +	if res, err := C.unlockpt(m); res < 0 {
>  		C.close(m)
>  		return nil, "", ptyError("unlockpt", err)
>  	}
> diff -Nru golang-1.19-1.19.6/src/runtime/mgcsweep.go golang-1.19-1.19.8/src/runtime/mgcsweep.go
> --- golang-1.19-1.19.6/src/runtime/mgcsweep.go	2023-02-14 01:38:47.000000000 +0800
> +++ golang-1.19-1.19.8/src/runtime/mgcsweep.go	2023-03-30 05:15:22.000000000 +0800
> @@ -815,11 +815,30 @@
>  		traceGCSweepStart()
>  	}
>  
> +	// Fix debt if necessary.
>  retry:
>  	sweptBasis := mheap_.pagesSweptBasis.Load()
> -
> -	// Fix debt if necessary.
> -	newHeapLive := uintptr(atomic.Load64(&gcController.heapLive)-mheap_.sweepHeapLiveBasis) + spanBytes
> +	live := atomic.Load64(&gcController.heapLive)
> +	liveBasis := mheap_.sweepHeapLiveBasis
> +	newHeapLive := spanBytes
> +	if liveBasis < live {
> +		// Only do this subtraction when we don't overflow. Otherwise, pagesTarget
> +		// might be computed as something really huge, causing us to get stuck
> +		// sweeping here until the next mark phase.
> +		//
> +		// Overflow can happen here if gcPaceSweeper is called concurrently with
> +		// sweeping (i.e. not during a STW, like it usually is) because this code
> +		// is intentionally racy. A concurrent call to gcPaceSweeper can happen
> +		// if a GC tuning parameter is modified and we read an older value of
> +		// heapLive than what was used to set the basis.
> +		//
> +		// This state should be transient, so it's fine to just let newHeapLive
> +		// be a relatively small number. We'll probably just skip this attempt to
> +		// sweep.
> +		//
> +		// See issue #57523.
> +		newHeapLive += uintptr(live - liveBasis)
> +	}
>  	pagesTarget := int64(mheap_.sweepPagesPerByte*float64(newHeapLive)) - int64(callerSweepPages)
>  	for pagesTarget > int64(mheap_.pagesSwept.Load()-sweptBasis) {
>  		if sweepone() == ^uintptr(0) {
> diff -Nru golang-1.19-1.19.6/src/sync/atomic/value.go golang-1.19-1.19.8/src/sync/atomic/value.go
> --- golang-1.19-1.19.6/src/sync/atomic/value.go	2023-02-14 01:38:48.000000000 +0800
> +++ golang-1.19-1.19.8/src/sync/atomic/value.go	2023-03-30 05:15:23.000000000 +0800
> @@ -190,5 +190,5 @@
>  }
>  
>  // Disable/enable preemption, implemented in runtime.
> -func runtime_procPin()
> +func runtime_procPin() int
>  func runtime_procUnpin()
> diff -Nru golang-1.19-1.19.6/src/time/zoneinfo.go golang-1.19-1.19.8/src/time/zoneinfo.go
> --- golang-1.19-1.19.6/src/time/zoneinfo.go	2023-02-14 01:38:48.000000000 +0800
> +++ golang-1.19-1.19.8/src/time/zoneinfo.go	2023-03-30 05:15:24.000000000 +0800
> @@ -182,7 +182,7 @@
>  	// If we're at the end of the known zone transitions,
>  	// try the extend string.
>  	if lo == len(tx)-1 && l.extend != "" {
> -		if ename, eoffset, estart, eend, eisDST, ok := tzset(l.extend, end, sec); ok {
> +		if ename, eoffset, estart, eend, eisDST, ok := tzset(l.extend, start, sec); ok {
>  			return ename, eoffset, estart, eend, eisDST
>  		}
>  	}
> @@ -243,12 +243,12 @@
>  }
>  
>  // tzset takes a timezone string like the one found in the TZ environment
> -// variable, the end of the last time zone transition expressed as seconds
> +// variable, the time of the last time zone transition expressed as seconds
>  // since January 1, 1970 00:00:00 UTC, and a time expressed the same way.
>  // We call this a tzset string since in C the function tzset reads TZ.
>  // The return values are as for lookup, plus ok which reports whether the
>  // parse succeeded.
> -func tzset(s string, initEnd, sec int64) (name string, offset int, start, end int64, isDST, ok bool) {
> +func tzset(s string, lastTxSec, sec int64) (name string, offset int, start, end int64, isDST, ok bool) {
>  	var (
>  		stdName, dstName     string
>  		stdOffset, dstOffset int
> @@ -269,7 +269,7 @@
>  
>  	if len(s) == 0 || s[0] == ',' {
>  		// No daylight savings time.
> -		return stdName, stdOffset, initEnd, omega, false, true
> +		return stdName, stdOffset, lastTxSec, omega, false, true
>  	}
>  
>  	dstName, s, ok = tzsetName(s)
> diff -Nru golang-1.19-1.19.6/src/time/zoneinfo_read.go golang-1.19-1.19.8/src/time/zoneinfo_read.go
> --- golang-1.19-1.19.6/src/time/zoneinfo_read.go	2023-02-14 01:38:48.000000000 +0800
> +++ golang-1.19-1.19.8/src/time/zoneinfo_read.go	2023-03-30 05:15:24.000000000 +0800
> @@ -329,7 +329,7 @@
>  			} else if l.extend != "" {
>  				// If we're at the end of the known zone transitions,
>  				// try the extend string.
> -				if name, offset, estart, eend, isDST, ok := tzset(l.extend, l.cacheEnd, sec); ok {
> +				if name, offset, estart, eend, isDST, ok := tzset(l.extend, l.cacheStart, sec); ok {
>  					l.cacheStart = estart
>  					l.cacheEnd = eend
>  					// Find the zone that is returned by tzset to avoid allocation if possible.
> diff -Nru golang-1.19-1.19.6/VERSION golang-1.19-1.19.8/VERSION
> --- golang-1.19-1.19.6/VERSION	2023-02-14 01:38:41.000000000 +0800
> +++ golang-1.19-1.19.8/VERSION	2023-03-30 05:15:14.000000000 +0800
> @@ -1 +1 @@
> -go1.19.6
> \ No newline at end of file
> +go1.19.8
> \ No newline at end of file


-- 
Sebastian Ramacher


Reply to: